Debian Codegraph @

ChangeLog

raw
1 2026-02-22 Tao Liu <[email protected]>
2 MACS 3.0.4
3
4 * Features added
5
6 1) `hmmratac` now supports `-f FRAG`, enabling direct processing of
7 single-cell fragment files.
8
9 Note: because FRAG uses a different internal subsampling
10 implementation from `BEDPE`/`BAMPE`, results can differ slightly
11 between collapsed `FRAG` input and standard paired-end inputs.
12
13 2) Added `--barcodes` and `--max-count` to `hmmratac` so peak
14 calling can be restricted to a barcode subset and/or capped by
15 fragment count.
16
17 For example, to call accessible regions for one cell type in an
18 scATAC-seq dataset: `--barcodes celltype1_barcodes.txt --max-count
19 2`.
20
21 * Bugs fixed
22
23 1) `PETrackII.pileup_bdg` now wraps pileup outputs in
24 `array.array` before passing them to `bedGraphTrackI`, preventing
25 runtime type errors when writing bedGraph tracks from FRAG inputs.
26
27 2) `PETrackII.sample_percent*` now correctly allows zero-percent
28 downsampling, preserving expected CLI behavior when the balance
29 target contains no fragments.
30
31 3) Fixed an overflow path in `PETrackI` when updating total track
32 length by explicitly casting `int32` values to `ulonglong`.
33
34 4) Fixed a `pvalue_stat` issue that could yield `nan` when a
35 region with the same p-value exceeded 2e9 bp (int32 limit).
36
37 * Documentation
38
39 1) Cleaned up source-code docstrings and expanded API docs for key
40 Signal and IO modules.
41
42 2) Added a Jupyter notebook demo showing single-cell ATAC-seq
43 processing with the MACS3 API.
44
45 2025-02-19 Tao Liu <[email protected]>
46 MACS 3.0.3
47
48 * Features added
49
50 1) Now support FRAG format for single-cell ATAC-seq in `callpeak`
51 and `pileup`. FRAG format is used by 10x Genomics to store
52 alignments from the single-cell ATAC-seq pipeline
53 `cellranger-atac` or the multiomics pipeline `cellranger-arc`. The
54 format is essentially BEDPE with two additional columns: the
55 barcode and the count of fragments aligned to the same location
56 with the same barcode. Support for FRAG in other tools is coming
57 soon, as well as for `hmmratac` calls.
58
59 If you specify `-f FRAG` as your input format:
60
61 - You can use a barcode list for a subset of cells with
62 `--barcodes`, then `callpeak` will identify peaks and `pileup`
63 will build pileup track for the fragments of this subset of cells.
64
65 - Duplicates will not get removed as we'll assume all fragments
66 are valid. Optionally, an option, `--max-count`, can be applied to
67 set the maximum count.
68
69 2) We transitioned our `pyx` codes to `py` codes, adopting a 'pure
70 Python style' with PEP-484 type annotations. This change has made
71 oursource codes more compatible with Python programming tools such
72 as `flake8`. During this process, we performed further code
73 cleaning and eliminated unnecessary dependencies. We intend to
74 continue improving our code quality in the future.
75
76 3) We have modified the handling of 'blacklist' regions in the
77 `hmmratac` tool. This change impacts both the
78 Expectation-Maximization (EM) step that estimates fragment length
79 distributions, and the Hidden Markov Model (HMM) step that learns
80 and predicts nucleosome states. We now exclude aligned fragments
81 located in the 'blocklist' regions before both steps. We
82 implemented the `exclude` functions in both PETrackI and PETrackII
83 to support this feature. For more detailed information and the
84 reasoning behind it, refer to issue #680.
85
86 4) We have tested Numpy>=2. Now MACS3 can be run on all Numpy >=
87 1.25.
88
89 * Bug fixed
90
91 1) The `hmmratagc` option `--keep-duplicate` previously had the
92 opposite effect of what its name and description
93 suggested. Therefore, it was renamed to `--remove-dup` to more
94 accurately describe the actual behavior. Duplicate fragments will
95 not be removed by `hmmratac` unless this option is explicitly set
96 up.
97
98 2) `hmmratac`: wrong class name was used while saving digested
99 signals in BedGraph files. Fixed multiple other issues related to
100 output filenames. #682
101
102 3) Fix issues in big-endian system in `Parser.py` codes. Enable
103 big-endian support in `BAM.py` codes for accessig certain
104 alignment records that overlap with givin genomic
105 coordinates using BAM/BAI files.
106
107 4) `predictd` and `filterdup`: wrong variable name used while
108 reading multiple pe/frag files.
109
110 * Doc
111
112 1) Explanation on the filtering criteria on SAM/BAM/BAMPE files.
113
114 2024-09-06 Tao Liu <[email protected]>
115 MACS 3.0.2
116
117 * Features added
118
119 1) Introduce a new emission model for the `hmmratac` function. Now
120 users can choose the simpler Poisson emission `--hmm-type poisson`
121 instead of the default Gaussian emission. As a consequence, the
122 saved HMM model file in json will include the hmm-type information
123 as well. Note that in order to be compatible with the HMM model
124 file from previous version, if there is no hmm-type information in
125 the model file, the hmm-type will be assigned as gaussian. #635
126
127 2) `hmmratac` now output narrowPeak format output. The summit
128 position and the peak score columns reported in the narrowPeak
129 output represents the position with highest foldchange value
130 (pileup vs average background).
131
132 3) Add `--cutoff-analysis-steps` and `--cutoff-analysis-max` for
133 `callpeak`, `bdgpeakcall`, and `hmmratac` so that we can
134 have finer resolution of the cutoff analysis report. #636 #642
135
136 4) Reduce memory usage of `hmmratac` during decoding step, by
137 writing decoding results to a temporary file on disk (file
138 location depends on the environmental TEMP setting), then loading
139 it back while identifying state pathes. This change will decrease
140 the memory usage dramatically. #628 #640
141
142 5) Fix instructions for preparing narrowPeak files for uploading
143 to UCSC browser, with the `--trackline` option in `callpeak`. #653
144
145 6) For gappedPeak output, set thickStart and thickEnd columns as
146 0, according to UCSC definition.
147
148 * Bugs fixed
149
150 1) Use `-O3` instead of `-Ofast` for compatibility. #637
151
152 * Documentation
153
154 1) Update instruction to install macs3 through conda/bioconda
155
156 2) Reorganize MACS3 docs and publish through
157 https://macs3-project.github.io/MACS
158
159 3) Description on various file formats used in MACS3.
160
161 2024-02-19 Tao Liu <[email protected]>
162 MACS 3.0.1
163
164 * Bugs fixed
165
166 1) Fixed a bug that the `hmmatac` can't correctly save the
167 digested signal files. #605 #611
168
169 2) Applied a patch to remove cython requirement from the installed
170 system. (it's needed for building the package). #606 #612
171
172 3) Relax the testing script while comparing the peaks called from
173 current codes and the standard peaks. To implement this, we added
174 'intersection' function to 'Regions' class to find the
175 intersecting regions of two Regions object (similar to PeakIO but
176 only recording chromosome, start and end positions). And we
177 updated the unit test 'test_Region.py' then implemented a script
178 'jaccard.py' to compute the Jaccard Index of two peak files. If
179 the JI > 0.99 we would think the peaks called and the standard
180 peaks are similar. This is to avoid the problem caused by
181 different Numpy/SciPy/sci-kit learn libraries, when certain peak
182 coordinates may have 10bps difference. #615 #619
183
184 4) Due to the changes in scikit-learn 1.3.0:
185 https://scikit-learn.org/1.3/whats_new/v1.3.html: The way hmmlearn
186 0.3 uses Kmeans will end up with inconsistent results between
187 sklearn <1.3 and sklearn >=1.3. Therefore, we patched the class
188 hmm.GaussianHMM and adjusted the standard output from `hmmratac`
189 subcommand. The change is based on
190 https://github.com/hmmlearn/hmmlearn/pull/545. The idea is to do
191 the random seeding of KMeans 10 times. Now the `hmmratac` results
192 should be more consistent (at least JI>0.99). #615 #620
193
194 * Other
195
196 1) We added some dependencies to MACS3. `hmmratc` subcommand needs
197 `hmmlearn` library, `hmmlearn` needs `scikit-learn` and
198 `scikit-learn` needs `scipy`. Since major releases have happened
199 for both`scipy` and `scikit-learn`, we have to set specific
200 version requirements for them in order to make sure the output
201 results from `hmmratac` are consistent.
202
203 2) We updated our documentation website using
204 Sphinx. https://macs3-project.github.io/MACS/
205
206 2023-11-15 Tao Liu <[email protected]>
207 MACS 3.0.0
208
209 1) Call variants in peak regions directly from BAM files. The
210 function was originally developed under code name SAPPER. Now
211 SAPPER has been merged into MACS as the `callvar` command. It can
212 be used to call SNVs and small INDELs directly from alignment
213 files for ChIP-seq or ATAC-seq. We call `fermi-lite` to assemble
214 the DNA sequence at the enriched genomic regions (binding sites or
215 accessible DNA) and to refine the alignment when necessary. We
216 added `simde` as a submodule in order to support fermi-lite
217 library under non-x64 architectures.
218
219 2) HMMRATAC module is added as subcommand `hmmratac`. HMMRATAC is
220 a dedicated software to analyze ATAC-seq data. The basic idea
221 behind HMMRATAC is to digest ATAC-seq data according to the
222 fragment length of read pairs into four signal tracks: short
223 fragments, mono-nucleosomal fragments, di-nucleosomal fragments
224 and tri-nucleosomal fragments. Then integrate the four tracks
225 again using Hidden Markov Model to consider three hidden states:
226 open region, nucleosomal region, and background region. The
227 orginal paper was published in 2019 written in JAVA, by Evan
228 Tarbell. We implemented it in Python/Cython and optimize the whole
229 process using existing MACS functions and hmmlearn. Now it can run
230 much faster than the original JAVA version. Note: evaluation of
231 the peak calling results is still underway.
232
233 3) Speed/memory optimization. Use the cykhash to replace python
234 dictionary. Use buffer (10MB) to read and parse input file (not
235 available for BAM file parser). And many optimization tweaks. We
236 added memory monitoring to the runtime messages.
237
238 4) R wrappers for MACS -- MACSr for bioconductor.
239
240 5) Code cleanup. Reorganize source codes.
241
242 6) Unit testing.
243
244 7) Switch to Github Action for CI, support multi-arch testing
245 including x64, armv7, aarch64, s390x and ppc64le. We also test on
246 Mac OS 12.
247
248 8) MACS tag-shifting model has been refined. Now it will use a
249 naive peak calling approach to find ALL possible paired peaks at +
250 and - strand, then use all of them to calculate the
251 cross-correlation. (a related bug has been fix
252 [#442](https://github.com/macs3-project/MACS/issues/442))
253
254 9) BAI index and random access to BAM file now is
255 supported. [#449](https://github.com/macs3-project/MACS/issues/449).
256
257 10) Support of Python > 3.10
258 [#498](https://github.com/macs3-project/MACS/issues/498)
259
260 11) The effective genome size parameters have been updated
261 according to
262 deeptools. [#508](https://github.com/macs3-project/MACS/issues/508)
263
264 12) Multiple updates regarding dependencies, anaconda built, CI/CD
265 process.
266
267 13) Cython 3 is supported.
268
269 14) Documentations for each subcommand can be found under /docs
270
271 *Other*
272
273 1) Missing header line while no peaks can be called
274 [#501](https://github.com/macs3-project/MACS/issues/501)
275 [#502](https://github.com/macs3-project/MACS/issues/502)
276
277 2) Note: different numpy, scipy, sklearn may give slightly
278 different results for hmmratac results. The current standard
279 results for automated testing in `/test` directory are from Numpy
280 1.25.1, Scipy 1.11.1, and sklearn 1.3.0.
281
282 2020-04-11 Tao Liu <[email protected]>
283 MACS version 2.2.7.1
284
285 * hotfix:
286
287 Add 'wheel' and 'pip' to pyproject.toml so that `pip install` can
288 work.
289
290 2020-04-10 Tao Liu <[email protected]>
291 MACS version 2.2.7
292
293 * Bugs fixed
294
295 1) MACS2 has been tested on multiple architectures to make sure it
296 can successfully generate consistent results. Currently the
297 supported architectures are: AMD64, ARM64, i386, PPC64LE, and
298 S390X. Thanks to @mr-c, @junaruga, and @tillea! Related to issue
299 #340, #349, #351, and #359; to PR #348, #350, #360, #361, #367,
300 and #370. The lesson is that if the project is built on Cython and
301 is aimed at memory efficiency, we should specifically define all
302 int/float types in pyx files such as int8_t or uint32_t using
303 either libc or numpy (c version) instead of relying on Cython
304 types such as short, long, double.
305
306 2) MACS2 setup script will check numpy and install numpy if
307 necessary. PR #378, issue #364
308
309 3) `bdgbroadcall` command will correctly add the score column (5th
310 column). The score (5th) column contains 10 times of the average
311 score in the broad region. PR #373, issue #362
312
313 4) The missing test on `bdgopt` subcommand has been added. PR #363
314
315 5) The obsolete option `--ratio` from `callpeak` subcommand has
316 been removed. PR #369, issue #366
317
318 6) Fixed the incorrect description in README on the 'maximum
319 length of broad region is 4 times of d' to 'maximum gap for
320 merging broad regions is 4 times of tag size by default'. PR #380,
321 issue #365.
322
323 * Other
324
325 1) CODE OF CONDUCT document has been added to MACS2 github
326 repository. PR #358
327
328 2019-12-12 Tao Liu <[email protected]>
329 MACS version 2.2.6
330
331 * New Features
332
333 1) Speed up MACS2. Some programming tricks and code cleanup. The
334 filter_dup function replaces separate_dups. The later one was
335 implemented for potentially putting back duplicate reads in
336 certain downstream analysis. However such analysis hasn't been
337 implemented. Optimize the speed of writing bedGraph
338 files. Optimize BAM and BAMPE parsing with pointer casting instead
339 of python unpack.
340
341 2) The comment lines in the headers of BED or SAM files will be
342 correctly skipped. However, MACS2 won't check comment lines in the
343 middle of the file.
344
345 * Bugs fixed
346
347 1) Cutoff-analysis in callpeak command. #341
348
349 2) Issues related to SAMParser and three ELAND Parsers are
350 fixed. #347
351
352 * Other
353
354 1) cmdlinetest script in test/ folder has been updated to: 1. test
355 cutoff-analysis with callpeak cmd; 2. output the 2 lines before
356 and after the error or warning message during tests; 3. output
357 only the first 10 lines if the difference between test result and
358 standard result can be found; 4. prockreport monitor CPU time and
359 memory usage in 1 sec interval -- a bit more accurate.
360
361 2) Python3.5 support is removed. Now MACS2 requires Python>=3.6.
362
363 2019-10-31 Tao Liu <[email protected]>
364 MACS version 2.2.5 (Py3 speed up)
365
366 * Features added
367
368 1) *Github code only and Not included in MACS2 release* New
369 testing data for performance test. An subsampled ENCODE2 CTCF
370 ChIP-seq dataset, including 5million ChIP reads and 5 million
371 control reads, has been included in the test folder for testing
372 CPU and memory usage (i.e. 5M test). Several related scripts ,
373 including `prockreport` for output cpu memory usage, `pyprofile`
374 and `pyprofile_stat` for debuging and profiling MACS2 codes, have
375 been included.
376
377 2) Speed up pvalue-qvalue checkup (pqtable checkup) #335 #338.
378 The old hashtable.pyx implementation copied from Pandas (very old
379 version) doesn't work well in Python3+Cython. It slows down the
380 pqtable checkup using the identical Cython codes as in
381 v2.1.4. While running 5M test, the `__getitem__` function in the
382 hashtable.pyx took 3.5s with 37,382,037 calls in MACS2 v2.1.4, but
383 148.6s with the same number of calls in MACS2 v2.2.4. As a
384 consequence, the standard python dictionary implementation has
385 replaced hashtable.pyx for pqtable checkup. Now MACS2 runs a bit
386 faster than py2 version, but uses a bit more memory. In general,
387 v2.2.5 can finish 5M reads test in 20% less time than MACS2
388 v2.1.4, but use 15% more memory.
389
390 * Bug fixed
391
392 1) More Python3 related fixes, e.g. the return value of keys from
393 py3 dict. #333 #337
394
395
396 2019-10-01 Tao Liu <[email protected]>
397 MACS version 2.2.4 (Python3)
398
399 * Features added
400
401 1) First Python3 version MACS2 released.
402
403 2) Version number 2.2.X will be used for MACS2 in Python3, in
404 parallel to 2.1.X.
405
406 3) More comprehensive test.sh script to check the consistency of
407 results from Python2 version and Python3 version.
408
409 4) Simplify setup.py script since the newest version transparently
410 supports cython. And when cython is not installed by the user,
411 setup.py can still compile using only C codes.
412
413 5) Fix Signal.pyx to use np.array instead of np.mat.
414
415 2019-09-30 Tao Liu <[email protected]>
416 MACS version 2.1.4
417
418 * Features added
419
420 Github Actions is used together with Travis CI for testing and
421 deployment.
422
423 * Bugs fixed
424
425 PR #322:
426
427 1) #318 Random score in bdgdiff output. It turns out the sum_v is
428 not initialized as 0 before adding. Potential bugs are fixed in
429 other functions in ScoreTrack and CallPeakUnit codes.
430
431 2) #321 Cython dependency in setup.py script is removed. And place
432 'cythonzie' call to the correct position.
433
434 3) A typo is fixed in Github Actions script.
435
436 2019-09-19 Tao Liu <[email protected]>
437 MACS version 2.1.3.3
438
439 * Features added
440
441 1) Support Docker auto-deploy. PR #309
442
443 2) Support Travis CI auto-testing, update unit-testing
444 scripts, and enable subcommand testing on small datasets.
445
446 3) Update README documents. #297 PR #306
447
448 4) `cmbreps` supports more than 2 replicates. Merged from PR #304
449 @Maarten-vd-Sande and PR #307 (our own chi-sq test code)
450
451 5) `--d-min` option is added in `callpeak` and `predictd`, to
452 exclude predictions of fragment size smaller than the given
453 value. Merged from PR #267 @shouldsee.
454
455 6) `--buffer-size` option is added in `predictd`, `filterdup`,
456 `pileup` and `refinepeak` subcommands. Users can use this option
457 to decrease memory usage while there are a large number of contigs
458 in the data. Also, now `callpeak`, `predictd`, `filterdup`,
459 `pileup` and `refinepeak` will suggest users to tweak
460 `--buffer-size` while catching a MemoryError. #313 PR #314
461
462 * Bugs fixed
463
464 1) #265 Fixed a bug where the pseudocount hasn't been applied
465 while calculating p-value score in ScoreTrack object.
466
467 2) Fixed bdgbroadcall so that it will report those broad peaks
468 without strong peak inside, a consistent behavior as `callpeak
469 --broad`.
470
471 3) Rename COPYING to LICENSE.
472
473 2018-10-17 Tao Liu <[email protected]>
474 MACS version 2.1.2
475
476 * New features
477
478 1) Added missing BEDPE support. And enable the support for BAMPE
479 and BEDPE formats in 'pileup', 'filterdup' and 'randsample'
480 subcommands. When format is BAMPE or BEDPE, The 'pileup' command
481 will pile up the whole fragment defined by mapping locations of
482 the left end and right end of each read pair. Thank @purcaro
483
484 2) Added options to callpeak command for tweaking max-gap and
485 min-len during peak calling. Thank @jsh58!
486
487 3) The callpeak option "--to-large" option is replaced with
488 "--scale-to large".
489
490 4) The randsample option "-t" has been replaced with "-i".
491
492 * Bug fixes
493
494 1) Fixed memory issue related to #122 and #146
495
496 2) Fixed a bug caused by a typo. Related to #249, Thank @shengqh
497
498 3) Fixed a bug while setting commandline qvalue cutoff.
499
500 4) Better describe the 5th column of narrowPeak. Thank @alexbarrera
501
502 5) Fixed the calculation of average fragment length for paired-end
503 data. Thank @jsh58
504
505 6) Fixed bugs caused by khash while computing p/q-value and log
506 likelihood ratios. Thank @jsh58
507
508 7) More spelling tweaks in source code. Thank @mr-c
509
510 2016-03-09 Tao Liu <[email protected]>
511 MACS version 2.1.1 20160309
512
513 * Retire the tag:rc.
514
515 * Fixed spelling. Merged pull request #120. Thank @mr-c!
516
517 * Change filtering criteria for reading BAM/SAM files
518
519 Related to callpeak and filterdup commands. Now the
520 reads/alignments flagged with 1028 or 'PCR/Optical duplicate' will
521 still be read although MACS2 may decide them as duplicates
522 later. Related to old issue #33. Sorry I forgot to address it for
523 years!
524
525 2016-02-26 Tao Liu <[email protected]>
526 MACS version 2.1.1 20160226 (tag:rc Zhengyue)
527
528 * Bug fixes
529
530 1) Now "-Ofast" has been replaced by "-O3 --ffast-math", because
531 the former option is not supported by older GCC. Related to issues
532 #91, #109.
533
534 2) Issue #108 is fixed. If no peak can be found in a chromosome,
535 the PeakIO won't throw an error.
536
537 * New features
538
539 1) callpeak
540
541 a) A more flexible format, BEDPE, is supported. Now users can
542 define the left and right position of the ChIPed fragment, and
543 MACS2 will skip model building and directly pileup the
544 fragments. Related to issue #112.
545
546 b) The 'tempdir' can be specified, to save cached pileup
547 tracks. Originially, the temporary files were stored in
548 /tmp. Thank @daler! Related to issues #97 and #105.
549
550 2) bdgopt
551
552 New operations are added, to calculate the maximum or minimum value between
553 values in BEDGRAPH and given value.
554
555 3) bdgcmp
556
557 New method is added, to calculate the maximum value between values
558 defined in two BEDGRAPH files.
559
560 2015-12-22 Tao Liu <[email protected]>
561 MACS version 2.1.0 20151222 (tag:rc Dongzhi)
562
563 * Bug fixes
564
565 1) Fix a bug while dealing with some chromosomes only containing
566 one read (pair). The size of dup_plus/dup_minus arrays after
567 filtering dups should +1.
568
569 2) Fix a bug related to the broad peak calling function in
570 previous versions. The gaps were miscalculated, so segmented weak
571 broad calls may be reported, and sometimes you would see peaks
572 with lower than cutoff values in the output files.
573
574 3) "Potentially" Fixed issue #105 on temporary cache files, need
575 further followup.
576
577
578 2015-07-31 Tao Liu <[email protected]>
579 MACS version 2.1.0 20150731 (tag:rc)
580
581 * Bug fixes
582
583 1) Fixed issue #76: information about broad/narrow cutoff will be
584 correctly displayed.
585
586 2) Fixed issue #79: bdgopt extparam option is fixed.
587
588 3) Fixed issue #87: reference to cProb has been fixed as 'Prob'
589 for filterdup command.
590
591 4) Fixed issue #78, #88 and similar issue reported in MACS google
592 group: MACS2 now can correctly deal with multiple alignment files
593 for -t or -c. The 'finalize' function will be correctly
594 called. Multiple files option is enabled for filterdup,
595 randsample, predictd, pileup and refinepeak commands.
596
597 5) A related issue to #88, when BAMPE mode is used, PE pairs will
598 be sorted by leftmost then rightmost ends.
599
600 6) Fixed issue #86: A wrong use of 'ndarray' to create Numpy
601 array. This will cause 'callpeak --nolambda' hang forever while
602 calculating pvalues and qvalues.
603
604 2015-04-20 Tao Liu <[email protected]>
605 MACS version 2.1.0 20150420 (tag:rc)
606
607 * New commands
608
609 1) bdgopt: some convenient functions to modify bedGraph files.
610
611 2) cmbreps: Combine scores from two replicates. Including three
612 methods: 1. take the maximum; 2. take the average; 3. use Fisher's
613 method to combine two p-value scores. After that, user can use
614 bdgpeakcall to call peaks on combined scores.
615
616 * New features
617
618 1) callpeak and bdgpeakcall now can try to analyze the
619 relationship between p-values and number/length of peaks then
620 generate a summary to help users decide an appropriate cutoff.
621
622 2) callpeak now can accept fold-enrichment cutoff as a filter for
623 final peak calls.
624
625 * Performance
626
627 Now MACS2 runs about 3X as fast as previous version. Trade
628 clean python codes for speed... Now while processing 50M ChIP vs
629 50M control, it will take only 10 minutes.
630
631 * Bug fixes
632
633 1) Sampling function in BAMPE mode.
634
635 2) Callpeak while there are >= 2 input files for -t or -c.
636
637 3) While reading BAM/SAM, those secondary or supplementary
638 alignments will be correctly skipped.
639
640 4) Fixed issue #33: Explanation is added to callpeak --keep-dup
641 option that MACS2 will discard those SAM/BAM alignments with bit
642 1024 no matter how --keep-dup is set.
643
644 5) Fixed issue #49: setuptools is used intead of distutils
645
646 6) Fixed issue #51: fix the problem when using --trackline
647 argument when control file is absent.
648
649 7) Fixed issue #53: Use Use SAM/BAM CIGAR to find the 5' end of
650 read mapped to minus strand. Previous implementation will find
651 incorrect 5' end if there is indel in alignment.
652
653 8) Fixed issue #56: An incorrect sorting method used for BAMPE
654 mode which will cause incorrect filtering of duplicated reads. Now
655 fixed.
656
657 9) Issue #63: Merged from jayhesselberth@github, extsize now can
658 be 1.
659
660 10) Issue #71: Merged from aertslab@github, close file descriptor
661 after creating them with mkstemp().
662
663 2014-06-16 Tao Liu <[email protected]>
664 MACS version 2.1.0 20140616 (tag:rc)
665
666 * callpeak module
667
668 "--ratio" is added to manually assign the scaling factor of ChIP
669 vs control, e.g. from NCIS. Thank Colin D and Dietmar Rieder for
670 implementing the patch file!
671
672 "--shift" is added to move cutting ends (5' end of reads) around,
673 in order to process DNAse-Seq data, e.g., use "--shift -100
674 --extsize 200" to get 200bps fragments around 5' ends. For general
675 ChIP-Seq data analysis, this option should be always set as
676 0. Thank Xi Chen and Anshul Kundaje for the discussions in user
677 group!
678
679 ** Do not output negative fragment size from cross-correlation
680 analysis. Thank Alvin Qin for the feedback!
681
682 ** --half-ext and --control-shift are removed. For complex read
683 shifting and extending, combine '--shift' and '--extsize'
684 options. For comparing two conditions, use 'bdgdiff' module
685 instead.
686
687 ** a bug is fixed to output the last pileup value in bdg file
688 correctly.
689
690 * filterdup
691
692 A 'dry-run' option is added to only output numbers, including the
693 number of allowed duplicates, the total number of reads before and
694 after filtering duplicates and the estimated duplication
695 rate. Thank John Urban for the suggestion!
696
697
698 2013-12-16 Tao Liu <[email protected]>
699 MACS version 2.0.10 20131216 (tag:alpha)
700
701 bug fixes and tweaks
702
703 * We changed license from Artistic License to 3-clauses BSD license.
704
705 Yes. Simpler the better.
706
707 * Process paired-end data with "-f BAMPE" without control
708
709 * GappedPeak output for --broad option has been fixed again to be
710 consistent with official UCSC format. We add 1bp pseudo-block to
711 left and/or right of broad region when necessary, so that you can
712 virtualize the regions without strong enrichment inside
713 successfully. In downstream analysis except for virtualization,
714 you may need to remove all 1bps blocks from gappedPeak file.
715
716 * diffpeak subcommand is temporarily disabled. Till we
717 re-implement it.
718
719 2013-10-28 Tao Liu <[email protected]>
720 MACS version 2.0.10 20131028 (tag:alpha)
721
722 * callpeak --call-summits improvement
723
724 The smoothing window length has been fixed as fragment length
725 instead of short read length. The larger smoothing window will
726 grant better smoothing results and better sub-peak summits
727 detection.
728
729 * --outdir and --ofile options for almost all commands
730
731 Thank Björn Grüning for initially implementing these options!
732 Now, MACS2 will save results into a specified
733 directory by '--outdir' option, and/or save result into a
734 specified file by '--ofile' option. Note, in case '--ofile' is
735 available for a subcommand, '-o' now has been adjusted to be the
736 same as '--ofile' instead of '--o-prefix'.
737
738 Here is the list of changes. For more detail, use 'macs2 xxx -h'
739 for each subcommand:
740
741 ** callpeak: --outdir
742 ** diffpeak: Not implemented
743 ** bdgpeakcall: --outdir and --ofile
744 ** bdgbroadcall: --outdir and --ofile
745 ** bdgcmp: --outdir and --ofile. While --ofile is used, the number
746 and the order of arguments for --ofile must be the same as for -m.
747 ** bdgdiff: --outdir and --ofile
748 ** filterdup: --outdir
749 ** pileup: --outdir
750 ** randsample: --outdir
751 ** refinepeak: --outdir and --ofile
752
753
754 2013-09-15 Tao Liu <[email protected]>
755 MACS version 2.0.10 20130915 (tag:alpha)
756
757 * callpeak Added a new option --buffer-size
758
759 This option is to tweak a previously hidden parameter that
760 controls the steps to increase array size for storing alignment
761 information. While in some rare cases, the number of
762 chromosomes/contigs/scaffolds is huge, the original default
763 setting will cause a huge memory waste. In these cases, we
764 recommend to decrease --buffer-size (e.g., 1000) to save memory,
765 although the decrease will slow process to read alignment files.
766
767 * an optimization to speed up pvalue-qvalue statistics
768
769 Previously, it took a hour to prepare p-q-table for 65M vs 65M
770 human TF library, and now it will take 10 minutes. It was due to a
771 single line of code to get a value from a numpy array ...
772
773 * fixed logLR bugs.
774
775 2013-07-31 Tao Liu <[email protected]>
776 MACS version 2.0.10 20130731 (tag:alpha)
777
778 * callpeak --call-summits
779
780 Fix bugs causing callpeak --call-summits option generating extra
781 number of peaks and inconsistent peak boundaries comparing to
782 default option. Thank Ben Levinson!
783
784 * bdgcmp output
785
786 Fix bugs causing bdgcmp output logLR all in positive values. Now
787 'depletion' can be correctly represented as negative values.
788
789 * bdgdiff
790
791 Fix the behavior of bdgdiff module. Now it can take four
792 bedGraph files, then use logLR as cutoff to call differential
793 regions. Check command line of bdgdiff for detail.
794
795 2013-07-13 Tao Liu <[email protected]>
796 MACS version 2.0.10 20130713 (tag:alpha)
797
798 * fix bugs while output broadPeak and gappedPeak.
799
800 Note. Those weak broad regions without any strong enrichment
801 regions inside won't be saved in gappedPeak file.
802
803 * bdgcmp -T and -C are merged into -S and description is updated.
804
805 Now, you can use it to override SPMR values in your input for
806 bdgcmp. To use SPMR (from 'callpeak --SPMR -B') while calculating
807 statistics will cause weird results ( in most cases, lower
808 significancy), and won't be consistent with MACS2 callpeak
809 behavior. So if you have SPMR bedGraphs, input the smaller/larger
810 sample size in MILLION according to 'callpeak --to-large' option.
811
812 2013-07-10 Tao Liu <[email protected]>
813 MACS version 2.0.10 20130710 (tag:alpha)
814
815 * fix BED style output format of callpeak module:
816
817 1) without --broad: narrowPeak (BED6+4) and BED for summit will be
818 the output. Old BED format file won't be saved.
819
820 2) with --broad: broadPeak (BED6+3) for broad region and
821 gappedPeak (BED12+3) for chained enriched regions will be the
822 output. Old BED format, narrowPeak format, summit file won't be
823 saved.
824
825 * bdgcmp now can accept list of methods to calculate scores. So
826 you can run it once to generate multiple types of scores. Thank
827 Jon Urban for this suggestion!
828
829 * C codes are re-generated through Cython 0.19.1.
830
831 2013-05-21 Tao Liu <[email protected]>
832 MACS version 2.0.10 20130520 (tag:alpha)
833
834 * broad peak calling modules are modified in order to report all
835 relexed regions even there is no strong enrichment inside.
836
837 2013-05-01 Tao Liu <[email protected]>
838 MACS version 2.0.10 20130501 (tag:alpha)
839
840 * Memory usage is decreased to about 1/4-1/5 of previous usage
841 Now, the internal data structure and algorithm are both
842 re-organized, so that intermediate data wouldn't be saved in
843 memory. Intead they will be calculated on the fly. New MACS2 will
844 spend longer time (1.5 to 2 times) however it will use less memory
845 so can be more usable on small mem servers.
846
847 * --seed option is added to callpeak and randsample commands
848 Thank Mathieu Gineste for this suggestion!
849
850 2013-03-05 Tao Liu <[email protected]>
851 MACS version 2.0.10 20130306 (tag:alpha)
852
853 * diffpeak module New module to detect differential binding sites
854 with more statistics.
855
856 * Introduced --refine-peaks
857 Calculates reads balancing to refine peak summits
858
859 * Ouput file names prefix
860 Correct encodePeak to narrowPeak, broadPeak to bed12.
861
862 2012-09-13 Benjamin Schiller <[email protected]>, Tao Liu <[email protected]>
863 MACS version 2.0.10 (tag:alpha not released)
864
865 * Introduced BAMPEParser
866 Reads PE data directly, requires bedtools for now
867
868 * Introduced --call-summits
869 Uses signal processing methods to call overlapping peaks
870
871 * Added --no-trackline
872 By default, files have descriptive tracklines now
873
874 * new refinepeak command (experimental)
875 This new function will use a similar method in SPP (wtd), to
876 analyze raw tag distribution in peak region, then redefine the
877 peak summit where plus and minus tags are evenly distributed
878 around.
879
880 * Changes to output *
881 cPeakDetect.pyx has full support for new print/write methods and
882 --call-peaks, BAMPEParser, and use of paired-end data
883
884 * Parser optimization
885
886 cParser.pyx is rewritten to use io.BufferedReader to speed
887 up. Speed is doubled.
888
889 Code is reorganized -- most of functions are inherited from
890 GenericParser class.
891
892 * Use cross-correlation to calculate fragment size
893
894 First, all pairs will be used in prediction for fragment
895 size. Previously, only no more than 1000 pairs are used. Second,
896 cross-correlation is used to find the best phase difference
897 between + and - tag pileups.
898
899 * Speed up p-value and q-value calculation
900
901 This part is ten times faster now. I am using a dictionary to
902 cache p-value results from Poisson CDF function. A bit more memory
903 will be used to increase speed. I hope this dictionary would not
904 explode since the possible pairs of ChIP signal and control lambda
905 are hugely redundant. Also, I rewrited part of q-value
906 calculation.
907
908 * Speed up peak detection
909
910 This part is about hundred of times faster now. Optimizations
911 include using Numpy functions as much as possible, and making loop
912 body as small as possible.
913
914 * Post-processing on differential calls
915
916 After macs2diff finds differential binding sites between two
917 conditions, it will try to annotate the peak calls from one of two
918 conditions, describe the changes ...
919
920 * Fragment size prediction in macs2diff
921
922 Now by default, macs2diff will try to use the average fragment
923 size from both condition 1 and condition 2 for tag extension and
924 peak calling. Previously, by default, it will use different sizes
925 unless --nomodel is specified.
926
927 Technically, I separate model building processes out. So macs2diff
928 will build fragment sizes for condition 1 and 2 in parallel (2
929 processes maximum), then perform 4-way comparisons in parallel (4
930 processes maximum).
931
932 * Diff score
933
934 Combine two p/qscore tracks together. At regions where condition 1
935 is higher than condition 2, score would be positive, otherwise,
936 negative.
937
938 * SAMParser and BAMParser
939
940 Bug fixed for paired-end sequencing data.
941
942 * BedGraph.pyx
943
944 Fixed a bug while calling peaks from BedGraph file. It previously
945 mistakenly output same peaks multiple times at the end of
946 chromosome.
947
948 2011-11-2 Tao Liu <[email protected]>
949 MACS version 2.0.9 (tag:alpha)
950
951 * Auto fixation on predicted d is turned off by default!
952
953 Previous --off-auto is now default. MACS will not automatically
954 fix d less than 2 times of tag size according to
955 --shiftsize. While tag size is getting longer nowadays, it would
956 be easier to have d less than 2 times of tag size, however d may
957 still be meaningful and useful. Please judge it using your own
958 wisdom.
959
960 * Scaling issue
961
962 Now, the default scaling while treatment and input are unbalanced
963 has been adjusted. By default, larger sample will be scaled down
964 linearly to match the smaller sample. In this way, background
965 noise will be reduced more than real signals, so we expect to have
966 more specific results than the other way around (i.e. --to-large
967 is set).
968
969 Also, an alternative option to randomly sample larger data
970 (--down-sample) is provided to replace default linear
971 scaling. However, this option will cause results irresproducible,
972 so be careful.
973
974 * randsample script
975
976 A new script 'randsample' is added, which can randomly sample
977 certain percentage or number of tags.
978
979 * Peak summit
980
981 Now, MACS will decide peak summits according to pileup height
982 instead of qvalue scores. In this way, the summit may be more
983 accurate.
984
985 * Diff score
986
987 MACS calculate qvalue scores as differential scores. When compare
988 two conditions (saying A and B), the maximum qscore for comparing
989 A to B -- maxqscore_a2b, and for comparing B to A --maxqscore_b2a
990 will be computed. If maxqscore_a2b is bigger, the diff score is
991 +maxqscore_a2b, otherwise, diff score is -1*maxqscore_b2a.
992
993 2011-09-15 Tao Liu <[email protected]>
994 MACS version 2.0.8 (tag:alpha)
995
996 * bin/macs2, bin/bdgbroadcall, MACS2/IO/cScoreTrack.pyx, MACS2/IO/cBedGraph.pyx
997
998 New script bdgbroadcall and the extra option '--broad' for macs2
999 script, can be used to call broad regions with a loose cutoff to
1000 link nearby significant regions. The output is represented as
1001 BED12 format.
1002
1003 * MACS2/IO/cScoreTrack.pyx
1004
1005 Fix q-value calculation to generate forcefully monotonic values.
1006
1007 * bin/eland*2bed, bin/sam2bed and bin/filterdup
1008
1009 They are combined to one more powerful script called
1010 "filterdup". The script filterdup can filter duplicated reads
1011 according to sequencing depth and genome size. The script can also
1012 convert any format supported by MACS to BED format.
1013
1014 2011-08-21 Tao Liu <[email protected]>
1015 MACS version 2.0.7 (tag:alpha)
1016
1017 * bin/macsdiff renamed to bin/bdgdiff
1018
1019 Now this script will work as a low-level finetuning tool as bdgcmp
1020 and bdgpeakcall.
1021
1022 * bin/macs2diff
1023
1024 A new script to take treatment and control files from two
1025 condition, calculate fragment size, use local poisson to get
1026 pvalues and BH process to get qvalues, then combine 4-ways result
1027 to call differential sites.
1028
1029 This script can use upto 4 cpus to speed up 4-ways calculation. (
1030 I am trying multiprocessing in python. )
1031
1032 * MACS2/Constants.py, MACS2/IO/cBedGraph.pyx,
1033 MACS2/IO/cScoreTrack.pyx, MACS2/OptValidator.py,
1034 MACS2/PeakModel.py, MACS2/cPeakDetect.pyx
1035
1036 All above files are modified for the new macs2diff script.
1037
1038 * bin/macs2, bin/macs2diff, MACS2/OptValidator.py
1039
1040 Now q-value 0.01 is the default cutoff. If -p is specified,
1041 p-value cutoff will be used instead.
1042
1043 2011-07-25 Tao Liu <[email protected]>
1044 MACS version 2.0.6 (tag:alpha)
1045
1046 * bin/macsdiff
1047
1048 A script to call differential regions. A naive way is introduced
1049 to find the regions where:
1050
1051 1. signal from condition 1 is larger than input 1 and condition 2 --
1052 unique region in condition 1;
1053 2. signal from condition 2 is larger than input 2 and condition 1
1054 -- unique region in condition 2;
1055 3. signal from condition 1 is larger than input 1, signal from
1056 condition 2 is larger than input 2, however either signal from
1057 condition 1 or 2 is not larger than the other.
1058
1059 Here 'larger' means the pvalue or qvalue from a Poisson test is
1060 under certain cutoff.
1061
1062 (I will make another script to wrap up mulitple scripts for
1063 differential calling)
1064
1065 2011-07-07 Tao Liu <[email protected]>
1066 MACS version 2.0.5 (tag:alpha)
1067
1068 * bin/macs2, MACS2/cPeakDetect.py, MACS2/IO/cScoreTrack.pyx,
1069 MACS2/IO/cPeakIO.pyx
1070
1071 Use hash to store peak information. Add back the feature to deal
1072 with data without control.
1073
1074 Fix bug which incorrectly allows small peaks at the end of
1075 chromosomes.
1076
1077 * bin/bdgpeakcall, bin/bdgcmp
1078
1079 Fix bugs. bdgpeakcall can output encodePeak format.
1080
1081 2011-06-22 Tao Liu <[email protected]>
1082 MACS version 2.0.4 (tag:alpha)
1083
1084 * cPeakDetect.py
1085
1086 Fix a bug, correctly assign lambda_bg while --to-small is
1087 set. Thanks Junya Seo!
1088
1089 Add rank and num of bp columns to pvalue-qvalue table.
1090
1091 * cScoreTrack.py
1092
1093 Fix bugs to correctly deal with peakless chromosomes. Thanks
1094 Vaibhav Jain!
1095
1096 Use AFDR for independent tests instead.
1097
1098 * encodePeak
1099
1100 Now MACS can output peak coordinates together with pvalue, qvalue,
1101 summit positions in a single encodePeak format (designed for
1102 ENCODE project) file. This file can be loaded to UCSC
1103 browser. Definition of some specific columns are: 5th:
1104 int(-log10pvalue*10), 7th: fold-change, 8th: -log10pvalue, 9th:
1105 -log10qvalue, 10th: relative summit position to peak start.
1106
1107
1108 2011-06-19 Tao Liu <[email protected]>
1109 MACS version 2.0.3 (tag:alpha)
1110
1111 * Rich output with qvalue, fold enrichment, and pileup height
1112
1113 Calculate q-values using a refined Benjamini–Hochberg–Yekutieli
1114 procedure:
1115
1116 http://en.wikipedia.org/wiki/False_discovery_rate#Dependent_tests
1117
1118 Now we have a similiar xls output file as before. The differences
1119 from previous file are:
1120
1121 1. Summit now is absolute summit, instead of relative summit
1122 position;
1123 2. 'Pileup' is previous 'tag' column. It's the extended fragment
1124 pileup at the peak summit;
1125 3. We now use '-log10(pvalue)' instead of '-10log10(pvalue)', so
1126 5.00 means 1e-5, simple and less confusing.
1127 4. FDR column becomes '-log10(qvalue)' column.
1128 5. The pileup, -log10pvalue, fold_enrichment and -log10qvalue are
1129 the values at the peak summit.
1130
1131 * Extra output files
1132
1133 NAME_pqtable.txt contains pvalue and qvalue relationships.
1134
1135 NAME_treat_pvalue.bdg and NAME_treat_qvalue.bdg store -log10pvalue
1136 and -log10qvalue scores in BedGraph format. Nearby regions with
1137 the same value are not merged.
1138
1139 * Separation of FeatIO.py
1140
1141 Its content has been divided into cPeakIO.pyx, cBedGraph.pyx, and
1142 cFixWidthTrack.pyx. A modified bedGraphTrackI class was
1143 implemented to store pileup, local lambda, pvalue, and qvalue
1144 alltogether in cScoreTrack.pyx.
1145
1146 * Experimental option --half-ext
1147
1148 Suggested by NPS algorithm, I added an experimental option
1149 --half-ext to let MACS only extends ChIP fragment around its
1150 middle point for only 1/2 d.
1151
1152 2011-06-12 Tao Liu <[email protected]>
1153 MACS version 2.0.2 (tag:alpha)
1154
1155 * macs2
1156
1157 Add an error check to see if there is no common chromosome names
1158 from treatment file and control file
1159
1160 * cPeakDetect.pyx, cFeatIO.pyx, cPileup.pyx
1161
1162 Reduce memory usage by removing deepcopy() calls.
1163
1164 * Modify README documents and others.
1165
1166 2011-05-19 Tao Liu <[email protected]>
1167 MACS Version 2.0.1 (tag:alpha)
1168
1169 * cPileup.pyx, cPeakDetect.pyx and peak calling process
1170
1171 Jie suggested me a brilliant simple method to pileup fragments
1172 into bedGraph track. It works extremely faster than the previous
1173 function, i.e, faster than MACS1.3 or MACS1.4. So I can include
1174 large local lambda calculation in MACSv2 now. Now I generate three
1175 bedGraphs for d-size local bias, slocal-size and llocal-size local
1176 bias, and calculate the maximum local bias as local lambda
1177 bedGraph track.
1178
1179 Minor: add_loc in bedGraphTrackI now can correctly merge the
1180 region with its preceding region if their value are the same.
1181
1182 * macs2
1183
1184 Add an option to shift control tags before extension. By default,
1185 control tags will be extended to both sides regardless of strand
1186 information.
1187
1188 2011-05-17 Tao Liu <[email protected]>
1189 MACS Version 2.0.0 (tag:alpha)
1190
1191 * Use bedGraph type to store data internally and externally.
1192
1193 We can have theoretically one-basepair resolution profiles. 10
1194 times smaller in filesize and even smaller after converting to
1195 bigWig for visualization.
1196
1197 * Peak calling process modified. Better peak boundary detection.
1198
1199 Extend ChIP tag to d, and pileup to have a ChIP bedGraph. Extend
1200 Control tag to d and 1,000bp, and pileup to two bedGraphs. (1000bp
1201 one will be averaged to d size) Then calculate the maximum value
1202 of these two tracks and a global background, to have a
1203 local-lambda bedGraph.
1204
1205 Use -10log10poisson_pvalue as scores to generate a score track
1206 before peak calling.
1207
1208 A general peak calling based on a score cutoff, min length of peak
1209 and max gap between nearby peaks.
1210
1211 * Option changes.
1212
1213 Wiggle file output is removed. Now we only support bedGraph
1214 output. The generation of bedGraph is highly recommended since it
1215 will not cost extra time. In other words, bedGraph generation is
1216 internally run even you don't want to save bedGraphs on disk, due
1217 to the peak calling algorithm in MACS v2.
1218
1219 * cProb.pyx
1220
1221 We now can calculate poisson pvalue in log space so that the score
1222 (-10*log10pvalue) will not have a upper limit of 3100 due to
1223 precision of float number.
1224
1225 * Cython is adopted to speed up Python code.
1226
1227 2011-02-28 Tao Liu <[email protected]>
1228 Small fixes
1229
1230 * Replaced with a newest WigTrackI class and fixed the wignorm script.
1231
1232 2011-02-21 Tao Liu <[email protected]>
1233 Version 1.4.0rc2 (Valentine)
1234
1235 * --single-wig option is renamed to --single-profile
1236
1237 * BedGraph output with --bdg or -B option.
1238
1239 The BedGraph output provides 1bp resolution fragment pileup
1240 profile. File size is smaller than wig file. This option can be
1241 combined with --single-profile option to produce a bedgraph file
1242 for the whole genome. This option can also make --space,
1243 --call-subpeaks invalid.
1244
1245 * Fix the description of --shiftsize to correctly state that the
1246 value is 1/2 d (fragment size).
1247
1248 * Fix a bug in the call to __filter_w_control_tags when control is
1249 not available.
1250
1251 * Fix a bug on --to-small option. Now it works as expected.
1252
1253 * Fix a bug while counting the tags in candidate peak region, an
1254 extra tag may be included. (Thanks to Jake Biesinger!)
1255
1256 * Fix the bug for the peaks extended outside of chromosome
1257 start. If the minus strand tag goes outside of chromosome start
1258 after extension of d, it will be thrown out.
1259
1260 * Post-process script for a combined wig file:
1261
1262 The "wignorm" command can be called after a full run of MACS14 as
1263 a postprocess. wignorm can calculate the local background from the
1264 control wig file from MACS14, then use either foldchange,
1265 -10*log10(pvalue) from possion test, or difference after asinh
1266 transformation as the score to build a single wig track to
1267 represent the binding strength. This script will take a
1268 significant long time to process.
1269
1270 * --wigextend has been obsoleted.
1271
1272 2010-09-21 Tao Liu <[email protected]>
1273 Version 1.4.0rc1 (Starry Sky)
1274
1275 * Duplicate reads option
1276
1277 --keep-dup behavior is changed. Now user can specify how many
1278 reads he/she wants to keep at the same genomic location. 'auto' to
1279 let MACS decide the number based on binomial distribution, 'all'
1280 to let MACS keep all reads.
1281
1282 * pvalue and FDR fixes (Thanks to Prof. Zhiping Weng)
1283
1284 By default, MACS will now scale the smaller dataset to the bigger
1285 dataset. For instance, if IP has 10 million reads, and Input has 5
1286 million, MACS will double the lambda value calculated from Input
1287 reads while calling BOTH the positive peaks and negative
1288 peaks. This will address the issue caused by unbalanced numbers of
1289 reads from IP and Input. If --to-small is turned on, MACS will
1290 scale the larger dataset to the smaller one. So from now on, if d
1291 is fixed, then the peaks from a MACS call for A vs B should be
1292 identical to the negative peaks from a B vs A.
1293
1294 2010-09-01 Tao Liu <[email protected]>
1295 Version 1.4.0beta (summer wishes)
1296
1297 * New features
1298
1299 ** Model building
1300
1301 The default behavior in the model building step is slightly
1302 changed. When MACS can't find enough pairs to build model
1303 (implemented in alpha version) or the modeled fragment length is
1304 less than 2 times of tag length (implemented in beta version),
1305 MACS will use 2 times of --shiftsize value as fragment length in
1306 the later analysis. --off-auto can turn off this default behavior.
1307
1308 ** Redundant tag filtering
1309
1310 The IO module is rewritten. The redundant tag filtering process
1311 becomes simpler and works as promise. The maximum allowed number
1312 of tags at the exact same location is calculated from the
1313 sequencing depth and genome size using a binomial distribution,
1314 for both TREAMENT and CONTROL separately. ( previously only
1315 TREATMENT is considered ) The exact same location means the same
1316 coordination and the same strand. Then MACS will only keep at most
1317 this number of tags at the exact same location in the following
1318 analysis. An option --keep-dup can let MACS skip the filtering and
1319 keep all the tags. However this may bring in a lot of sequencing
1320 bias, so you may get many false positive peaks.
1321
1322 ** Single wiggle mode
1323
1324 First thing to mention, this is not the score track that I
1325 described before. By default, MACS generates wiggle files for
1326 fragment pileup for every chromosomes separately. When you use
1327 --single-wig option, MACS will generate a single wiggle file for
1328 all the chromosomes so you will get a wig.gz for TREATMENT and
1329 another wig.gz for CONTROL if available.
1330
1331 ** Sniff -- automatic format detection
1332
1333 Now, by default or "-f AUTO", MACS will decide the input file
1334 format automatically. Technically, it will try to read at most
1335 1000 records for the first 10 non-comment lines. If it succeeds,
1336 the format is decided. I recommend not to use AUTO and specify the
1337 right format for your input files, unless you combine different
1338 formats in a single MACS run.
1339
1340 * Options changes
1341
1342 --single-wig and --keep-dup are added. Check previous section in
1343 ChangeLog for detail.
1344
1345 -f (--format) AUTO is now the default option.
1346
1347 --slocal default: 1000
1348 --llocal default: 10000
1349
1350 * Bug fixed
1351
1352 Setup script will stop the installation if python version is not
1353 python2.6 or python2.7.
1354
1355 Local lambda calculation has been changed back. MACS will check
1356 peak_region, slocal( default 1K) and llocal (default 10K) for the
1357 local bias. The previous 200bps default will cause MACS misses
1358 some peaks where the input bias is very sharp.
1359
1360 sam2bed.py script is corrected.
1361
1362 Relative pos in xls output is fixed.
1363
1364 Parser for ELAND_export is fixed to pass some of the no match
1365 lines. And elandexport2bed.py is fixed too. ( however I can't
1366 guarantee that it works on any eland_export files. )
1367
1368 2010-06-04 Tao Liu <[email protected]>
1369 Version 1.4.0alpha2 (be smarter)
1370
1371 * Options changes
1372
1373 --gsize now provides shortcuts for common genomes, including
1374 human, mouse, C. elegans and fruitfly.
1375
1376 --llocal now will be 5000 bps if there is no input file, so that
1377 local lambda doesn't overkill enriched binding sites.
1378
1379 2010-06-02 Tao Liu <[email protected]>
1380 Version 1.4alpha (be smarter)
1381
1382 * Options changes
1383
1384 --tsize option is redesigned. MACS will use the first 10 lines of
1385 the input to decide the tag size. If user specifies --tsize, it
1386 will override the auto decided tsize.
1387
1388 --lambdaset is replaced by --slocal and --llocal which mean the
1389 small local region and large local region.
1390
1391 --bw has no effect on the scan-window size now. It only affects the
1392 paired-peaks model process.
1393
1394 * Model building
1395
1396 During the model building, MACS will pick out the enriched regions
1397 which are not too high and not too low to build the paired-peak
1398 model. Default the region is from fold 10 to fold 30. If MACS
1399 fails to build the model, by default it will use the nomodel
1400 settings, like shiftsize=100bps, to shift and extend each
1401 tags. This behavior can be turned off by '--off-auto'.
1402
1403 * Output files
1404
1405 An extra file including all the summit positions are saved in
1406 *_summits.bed file. An option '--call-subpeaks' will invoke
1407 PeakSplitter developed by Mali Salmon to split wide peaks into
1408 smaller subpeaks.
1409
1410 * Sniff ( will in beta )
1411
1412 Automatically recognize the input file format, so use can combine
1413 different format in one MACS run.
1414
1415 Not implemented features/TODO:
1416
1417 * Algorithms ( in near future? )
1418
1419 MACS will try to refine the peak boundaries by calculating the
1420 scores for every point in the candidate peak regions. The score
1421 will be the -10*log(10,pvalue) on a local poisson distribution. A
1422 cutoff specified by users (--pvalue) will be applied to find the
1423 precise sub-peaks in the original candidate peak region. Peak
1424 boudaries and peak summits positions will be saved in separate BED
1425 files.
1426
1427 * Single wiggle track ( in near future? )
1428
1429 A single wiggle track will be generated to save the scores within
1430 candidate peak regions in the 10bps resolution. The wiggle file
1431 is in fixedStep format.
1432
1433
1434 2009-10-16 Tao Liu <[email protected]>
1435 Version 1.3.7.1 (Oktoberfest, bug fixed #1)
1436
1437 * bin/Constants.py
1438
1439 Fixed typo. FCSTEP -> FESTEP
1440
1441 * lib/PeakDetect.py
1442
1443 The 'femax' attribute bug is fixed
1444
1445 2009-10-02 Tao Liu <[email protected]>
1446 Version 1.3.7 (Oktoberfest)
1447
1448 * bin/macs, lib/PeakDetect.py, lib/IO/__init__.py, lib/OptValidator.py
1449
1450 Enhancements by Peter Chines:
1451
1452 1. gzip files are supported.
1453 2. when --diag is on, user can set the increment and endpoint for
1454 fold enrichment analysis by setting --fe-step and --fe-max.
1455
1456 Enhancements by Davide Cittaro:
1457
1458 1. BAM and SAM formats are supported.
1459 2. small changes in the header lines of wiggle output.
1460
1461 Enhancements by Me:
1462 1. I added --fe-min option;
1463 2. Bowtie ascii output with suffix ".map" is supported.
1464
1465 Bug fixed:
1466
1467 1. --nolambda bug is fixed. ( reported by Martin in JHU )
1468 2. --diag bug is fixed. ( reported by Bogdan Tanasa )
1469 3. Function to remove suffix '.fa' is fixed. ( reported by Jeff Johnston )
1470 4. Some "fold change" have been changed to "fold enrichment".
1471
1472 2009-06-10 Tao Liu <[email protected]>
1473 Version 1.3.6.1 (default parameter change)
1474
1475 * bin/macs, lib/PeakDetect.py
1476
1477 "--oldfdr" is removed. The 'oldfdr' behaviour becomes
1478 default. "--futurefdr" is added which can turn on the 'new' method
1479 introduced in 1.3.6. By default it's off.
1480
1481 * lib/PeakDetect.py
1482
1483 Fixed a bug. p-value is corrected a little bit.
1484
1485
1486 2009-05-11 Tao Liu <[email protected]>
1487 Version 1.3.6 (Birthday cake)
1488
1489 * bin/macs
1490
1491 "track name" is added to the header of BED output file.
1492
1493 Now the default peak detection method is to consider 5k and 10k
1494 nearby regions in treatment data and peak location, 1k, 5k, and
1495 10k regions in control data to calculate local bias. The old
1496 method can be called through '--old' option.
1497
1498 Information about how many total/unique tags in treatment or
1499 control will be saved in final .xls output.
1500
1501 * lib/IO/__init__.py
1502
1503 ".fa" will be removed from input tag alignment so only the
1504 chromosome names are kept.
1505
1506 WigTrackI class is added for Wiggle like data structure. (not used
1507 now)
1508
1509 The parser for ELAND multi PET files has been fixed. Now the 5'
1510 tag position for a pair will be kept, whereas in the previous
1511 version, the middle points are kept.
1512
1513 * lib/IO/BinKeeper.py
1514
1515 BinKeeperI class is inspired by Jim Kent's library for UCSC genome
1516 browser, which can quickly access certain region for values in a
1517 large wiggle like data file. (not used now)
1518
1519 * lib/OptValidator.py
1520
1521 typo fixed.
1522
1523 * lib/PeakDetect.py
1524
1525 Now the default peak detection method is to consider 5k and 10k
1526 nearby regions in treatment data and peak location, 1k, 5k, and
1527 10k regions in control data to calculate local bias. The old
1528 method can be called through '--old' option.
1529
1530 Two columns have beed added to BED output file. 4th column: peak
1531 name; 5th column: peak score using -10log(10,pvalue) as score.
1532
1533 * setup.py
1534
1535 Add support to build a Mac App through 'setup.py py2app', or a
1536 Windows executable through 'setup.py py2exe'. You need to install
1537 py2app or py2exe package in order to use these functions.
1538
1539 2009-02-12 Tao Liu <[email protected]>
1540 Version 1.3.5 (local lambda fixed, typo fixed, model figure improved)
1541
1542 * PeakDetect.py
1543
1544 Now, besides 1k, 5k, 10k, MACS will also consider peak size region
1545 in control data to calculate local lambda for each peak. Peak
1546 calling results will be slightly different with previous version,
1547 beware!
1548
1549 * OptValidator.py
1550
1551 Typo fixed, ELANDParser -> ELANDResultParser
1552
1553 * OutputWriter.py
1554
1555 Now, modeled d value will be shown on the model figure.
1556
1557 2009-01-06 Tao Liu <[email protected]>
1558 Version 1.3.4 (Happy New Year Version, bug fixed, ELAND multi/PET support)
1559
1560 * macs, IO/__init__.py, PeakDetect.py
1561
1562 Add support for ELAND multi format. Add support for Pair-End
1563 experiment, in this case, 5'end and 3'end ELAND multi format files
1564 are required for treatment or control data. See 00README file for
1565 detail.
1566
1567 Add wigextend option.
1568
1569 Add petdist option for Pair-End Tag experiment, which is the best
1570 distance between 5' and 3' tags.
1571
1572 * PeakDetect.py
1573
1574 Fixed a bug which cause the end positions of every peak region
1575 incorrectly added by 1 bp. ( Thanks Mali Salmon!)
1576
1577 * OutputWriter.py
1578
1579 Fix bugs while generating wiggle files. The start position of
1580 wiggle file is set to 1 instead of 0.
1581
1582 Fix a bug that every 10M bps, signals in the first 'd' range are
1583 lower than actual. ( Thanks Mali Salmon!)
1584
1585
1586 2008-12-03 Tao Liu <[email protected]>
1587 Version 1.3.3 (wiggle bugs fixed)
1588
1589 * OutputWriter.py
1590
1591 Fix bugs while generating wiggle files. 1. 'span=' is added to
1592 'variableStep' line; 2. previously, every 10M bps, the coordinates
1593 were wrongly shifted to the right for 'd' basepairs.
1594
1595 * macs, PeakDetect.py
1596
1597 Add an option to save wiggle files on different resolution.
1598
1599 2008-10-02 Tao Liu <[email protected]>
1600 Version 1.3.2 (tiny bugs fixed)
1601
1602 * IO/__init__.py
1603
1604 Fix 65536 -> 65535. ( Thank Joon)
1605
1606 * Prob.py
1607
1608 Improved for binomial function with extra large number. Imported
1609 from Cistrome project.
1610
1611 * PeakDetect.py
1612
1613 If treatment channel misses reads in some chromosome included in
1614 control channel, or vice versa, MACS will not exit. (Thank Shaun
1615 Mahony)
1616
1617 Instead, MACS will fake a tag at position -1 when calling
1618 treatment peaks vs control, but will ignore the chromosome while
1619 calling negative peaks.
1620
1621 2008-09-04 Tao Liu <[email protected]>
1622 Version 1.3.1 (tiny bugs fixed version)
1623
1624 * Prob.py
1625
1626 Hyunjin Gene Shin contributed some codes to Prob.py. Now the
1627 binomial functions can tolerate large and small numbers.
1628
1629 * IO/__init__.py
1630
1631 Parsers now split lines in BED/ELAND file using any
1632 whitespaces. 'track' or 'browser' lines will be regarded as
1633 comment lines. A bug fixed when throwing StrandFormatError. The
1634 maximum redundant tag number at a single position can be no less
1635 than 65536.
1636
1637
1638 2008-07-15 Tao Liu <[email protected]>
1639 Version 1.3 (naming clarification version)
1640
1641 * Naming clarification changes according to our manuscript:
1642
1643 'frag_len' is changed to 'd'.
1644
1645 'fold_change' is changed to 'fold_enrichment'.
1646
1647 Suggest '--bw' parameter to be determined by users from the real
1648 sonication size.
1649
1650 Maximum FDR is 100% in the output file.
1651
1652 And other clarifications in 00README file and the documents on the
1653 website.
1654
1655 * IO/__init__.py
1656 If the redundant tag number at a single position is over 32767,
1657 just remember 32767, instead of raising an overflow exception.
1658
1659 * setup.py
1660 fixed a typo.
1661
1662 * PeakDetect.py
1663 Bug fixed for diagnosis report.
1664
1665
1666 2008-07-10 Tao Liu <[email protected]>
1667 Version 1.2.2gamma
1668
1669 * Serious bugs fix:
1670
1671 Poisson distribution CDF and inverse CDF functions are
1672 corrected. They can produce right results even for huge lambda
1673 now. So that the p-value and FDR values in the final excel sheet
1674 are corrected.
1675
1676 IO package now can tolerate some rare cases; ELANDParser in IO
1677 package is fixed. (Thank Bogdan)
1678
1679 * Improvement:
1680
1681 Reverse paired peaks in model are rejected. So there will be no
1682 negative 'frag_len'. (Thank Bogdan)
1683
1684 * Features added:
1685
1686 Diagnosis function is completed. Which can output a table file for
1687 users to estimate their sequencing depth.
1688
1689
1690 2008-06-30 Tao Liu <[email protected]>
1691 Version 1.2
1692
1693 * Probe.py is added!
1694
1695 GSL is totally removed from MACS. Instead, I have implemented the
1696 CDF and inverse CDF for poisson and binomial distribution purely
1697 in python.
1698
1699 * Constants.py is added!
1700
1701 Organize constants used in MACS in the Constants.py file.
1702
1703 * All other files are modified!
1704
1705 Foldchange calculation is modified. Now the foldchange only be
1706 calculated at the peak summit position instead of the whole peak
1707 region. The values will be higher and more robust than before.
1708
1709 Features added:
1710
1711 1. MACS can save wiggle format files containing the tag number at
1712 every 10 bp along the genome. Tags are shifted according to our
1713 model before they are calculated.
1714
1715 2. Model building and local lambda calculation can be skipped with
1716 certain options.
1717
1718 3. A diagnosis report can be generated through '--diag'
1719 option. This report can help you get an assumption about the
1720 sequencing saturation. This funtion is only in beta stage.
1721
1722 4. FDR calculation speed is highly improved.
1723
1724 2008-05-28 Tao Liu <[email protected]>
1725 Version 1.1
1726
1727 * TabIO, PeakModel.py ...
1728 Bug fixed to let MACS tolerate some cases while there is no tag on
1729 either plus strand or minus strand.
1730
1731 * setup.py
1732 Check the version of python. If the version is lower than 2.4,
1733 refuse to install with warning.
1734
1735
1736 2013-07-31 Tao Liu <[email protected]>
1737 MACS version 2.0.10 20130731 (tag:alpha)
1738
1739 * callpeak --call-summits
1740
1741 Fix bugs causing callpeak --call-summits option generating extra
1742 number of peaks and inconsistent peak boundaries comparing to
1743 default option. Thank Ben Levinson!
1744
1745 * bdgcmp output
1746
1747 Fix bugs causing bdgcmp output logLR all in positive values. Now
1748 'depletion' can be correctly represented as negative values.
1749
1750 * bdgdiff
1751
1752 Fix the behavior of bdgdiff module. Now it can take four
1753 bedGraph files, then use logLR as cutoff to call differential
1754 regions. Check command line of bdgdiff for detail.
1755
1756 2013-07-13 Tao Liu <[email protected]>
1757 MACS version 2.0.10 20130713 (tag:alpha)
1758
1759 * fix bugs while output broadPeak and gappedPeak.
1760
1761 Note. Those weak broad regions without any strong enrichment
1762 regions inside won't be saved in gappedPeak file.
1763
1764 * bdgcmp -T and -C are merged into -S and description is updated.
1765
1766 Now, you can use it to override SPMR values in your input for
1767 bdgcmp. To use SPMR (from 'callpeak --SPMR -B') while calculating
1768 statistics will cause weird results ( in most cases, lower
1769 significancy), and won't be consistent with MACS2 callpeak
1770 behavior. So if you have SPMR bedGraphs, input the smaller/larger
1771 sample size in MILLION according to 'callpeak --to-large' option.
1772
1773 2013-07-10 Tao Liu <[email protected]>
1774 MACS version 2.0.10 20130710 (tag:alpha)
1775
1776 * fix BED style output format of callpeak module:
1777
1778 1) without --broad: narrowPeak (BED6+4) and BED for summit will be
1779 the output. Old BED format file won't be saved.
1780
1781 2) with --broad: broadPeak (BED6+3) for broad region and
1782 gappedPeak (BED12+3) for chained enriched regions will be the
1783 output. Old BED format, narrowPeak format, summit file won't be
1784 saved.
1785
1786 * bdgcmp now can accept list of methods to calculate scores. So
1787 you can run it once to generate multiple types of scores. Thank
1788 Jon Urban for this suggestion!
1789
1790 * C codes are re-generated through Cython 0.19.1.
1791
1792 2013-05-21 Tao Liu <[email protected]>
1793 MACS version 2.0.10 20130520 (tag:alpha)
1794
1795 * broad peak calling modules are modified in order to report all
1796 relexed regions even there is no strong enrichment inside.
1797
1798 2013-05-01 Tao Liu <[email protected]>
1799 MACS version 2.0.10 20130501 (tag:alpha)
1800
1801 * Memory usage is decreased to about 1/4-1/5 of previous usage
1802 Now, the internal data structure and algorithm are both
1803 re-organized, so that intermediate data wouldn't be saved in
1804 memory. Intead they will be calculated on the fly. New MACS2 will
1805 spend longer time (1.5 to 2 times) however it will use less memory
1806 so can be more usable on small mem servers.
1807
1808 * --seed option is added to callpeak and randsample commands
1809 Thank Mathieu Gineste for this suggestion!
1810
1811 2013-03-05 Tao Liu <[email protected]>
1812 MACS version 2.0.10 20130306 (tag:alpha)
1813
1814 * diffpeak module New module to detect differential binding sites
1815 with more statistics.
1816
1817 * Introduced --refine-peaks
1818 Calculates reads balancing to refine peak summits
1819
1820 * Ouput file names prefix
1821 Correct encodePeak to narrowPeak, broadPeak to bed12.
1822
1823 2012-09-13 Benjamin Schiller <[email protected]>, Tao Liu <[email protected]>
1824 MACS version 2.0.10 (tag:alpha not released)
1825
1826 * Introduced BAMPEParser
1827 Reads PE data directly, requires bedtools for now
1828
1829 * Introduced --call-summits
1830 Uses signal processing methods to call overlapping peaks
1831
1832 * Added --no-trackline
1833 By default, files have descriptive tracklines now
1834
1835 * new refinepeak command (experimental)
1836 This new function will use a similar method in SPP (wtd), to
1837 analyze raw tag distribution in peak region, then redefine the
1838 peak summit where plus and minus tags are evenly distributed
1839 around.
1840
1841 * Changes to output *
1842 cPeakDetect.pyx has full support for new print/write methods and
1843 --call-peaks, BAMPEParser, and use of paired-end data
1844
1845 * Parser optimization
1846
1847 cParser.pyx is rewritten to use io.BufferedReader to speed
1848 up. Speed is doubled.
1849
1850 Code is reorganized -- most of functions are inherited from
1851 GenericParser class.
1852
1853 * Use cross-correlation to calculate fragment size
1854
1855 First, all pairs will be used in prediction for fragment
1856 size. Previously, only no more than 1000 pairs are used. Second,
1857 cross-correlation is used to find the best phase difference
1858 between + and - tag pileups.
1859
1860 * Speed up p-value and q-value calculation
1861
1862 This part is ten times faster now. I am using a dictionary to
1863 cache p-value results from Poisson CDF function. A bit more memory
1864 will be used to increase speed. I hope this dictionary would not
1865 explode since the possible pairs of ChIP signal and control lambda
1866 are hugely redundant. Also, I rewrited part of q-value
1867 calculation.
1868
1869 * Speed up peak detection
1870
1871 This part is about hundred of times faster now. Optimizations
1872 include using Numpy functions as much as possible, and making loop
1873 body as small as possible.
1874
1875 * Post-processing on differential calls
1876
1877 After macs2diff finds differential binding sites between two
1878 conditions, it will try to annotate the peak calls from one of two
1879 conditions, describe the changes ...
1880
1881 * Fragment size prediction in macs2diff
1882
1883 Now by default, macs2diff will try to use the average fragment
1884 size from both condition 1 and condition 2 for tag extension and
1885 peak calling. Previously, by default, it will use different sizes
1886 unless --nomodel is specified.
1887
1888 Technically, I separate model building processes out. So macs2diff
1889 will build fragment sizes for condition 1 and 2 in parallel (2
1890 processes maximum), then perform 4-way comparisons in parallel (4
1891 processes maximum).
1892
1893 * Diff score
1894
1895 Combine two p/qscore tracks together. At regions where condition 1
1896 is higher than condition 2, score would be positive, otherwise,
1897 negative.
1898
1899 * SAMParser and BAMParser
1900
1901 Bug fixed for paired-end sequencing data.
1902
1903 * BedGraph.pyx
1904
1905 Fixed a bug while calling peaks from BedGraph file. It previously
1906 mistakenly output same peaks multiple times at the end of
1907 chromosome.
1908
1909 2011-11-2 Tao Liu <[email protected]>
1910 MACS version 2.0.9 (tag:alpha)
1911
1912 * Auto fixation on predicted d is turned off by default!
1913
1914 Previous --off-auto is now default. MACS will not automatically
1915 fix d less than 2 times of tag size according to
1916 --shiftsize. While tag size is getting longer nowadays, it would
1917 be easier to have d less than 2 times of tag size, however d may
1918 still be meaningful and useful. Please judge it using your own
1919 wisdom.
1920
1921 * Scaling issue
1922
1923 Now, the default scaling while treatment and input are unbalanced
1924 has been adjusted. By default, larger sample will be scaled down
1925 linearly to match the smaller sample. In this way, background
1926 noise will be reduced more than real signals, so we expect to have
1927 more specific results than the other way around (i.e. --to-large
1928 is set).
1929
1930 Also, an alternative option to randomly sample larger data
1931 (--down-sample) is provided to replace default linear
1932 scaling. However, this option will cause results irresproducible,
1933 so be careful.
1934
1935 * randsample script
1936
1937 A new script 'randsample' is added, which can randomly sample
1938 certain percentage or number of tags.
1939
1940 * Peak summit
1941
1942 Now, MACS will decide peak summits according to pileup height
1943 instead of qvalue scores. In this way, the summit may be more
1944 accurate.
1945
1946 * Diff score
1947
1948 MACS calculate qvalue scores as differential scores. When compare
1949 two conditions (saying A and B), the maximum qscore for comparing
1950 A to B -- maxqscore_a2b, and for comparing B to A --maxqscore_b2a
1951 will be computed. If maxqscore_a2b is bigger, the diff score is
1952 +maxqscore_a2b, otherwise, diff score is -1*maxqscore_b2a.
1953
1954 2011-09-15 Tao Liu <[email protected]>
1955 MACS version 2.0.8 (tag:alpha)
1956
1957 * bin/macs2, bin/bdgbroadcall, MACS2/IO/cScoreTrack.pyx, MACS2/IO/cBedGraph.pyx
1958
1959 New script bdgbroadcall and the extra option '--broad' for macs2
1960 script, can be used to call broad regions with a loose cutoff to
1961 link nearby significant regions. The output is represented as
1962 BED12 format.
1963
1964 * MACS2/IO/cScoreTrack.pyx
1965
1966 Fix q-value calculation to generate forcefully monotonic values.
1967
1968 * bin/eland*2bed, bin/sam2bed and bin/filterdup
1969
1970 They are combined to one more powerful script called
1971 "filterdup". The script filterdup can filter duplicated reads
1972 according to sequencing depth and genome size. The script can also
1973 convert any format supported by MACS to BED format.
1974
1975 2011-08-21 Tao Liu <[email protected]>
1976 MACS version 2.0.7 (tag:alpha)
1977
1978 * bin/macsdiff renamed to bin/bdgdiff
1979
1980 Now this script will work as a low-level finetuning tool as bdgcmp
1981 and bdgpeakcall.
1982
1983 * bin/macs2diff
1984
1985 A new script to take treatment and control files from two
1986 condition, calculate fragment size, use local poisson to get
1987 pvalues and BH process to get qvalues, then combine 4-ways result
1988 to call differential sites.
1989
1990 This script can use upto 4 cpus to speed up 4-ways calculation. (
1991 I am trying multiprocessing in python. )
1992
1993 * MACS2/Constants.py, MACS2/IO/cBedGraph.pyx,
1994 MACS2/IO/cScoreTrack.pyx, MACS2/OptValidator.py,
1995 MACS2/PeakModel.py, MACS2/cPeakDetect.pyx
1996
1997 All above files are modified for the new macs2diff script.
1998
1999 * bin/macs2, bin/macs2diff, MACS2/OptValidator.py
2000
2001 Now q-value 0.01 is the default cutoff. If -p is specified,
2002 p-value cutoff will be used instead.
2003
2004 2011-07-25 Tao Liu <[email protected]>
2005 MACS version 2.0.6 (tag:alpha)
2006
2007 * bin/macsdiff
2008
2009 A script to call differential regions. A naive way is introduced
2010 to find the regions where:
2011
2012 1. signal from condition 1 is larger than input 1 and condition 2 --
2013 unique region in condition 1;
2014 2. signal from condition 2 is larger than input 2 and condition 1
2015 -- unique region in condition 2;
2016 3. signal from condition 1 is larger than input 1, signal from
2017 condition 2 is larger than input 2, however either signal from
2018 condition 1 or 2 is not larger than the other.
2019
2020 Here 'larger' means the pvalue or qvalue from a Poisson test is
2021 under certain cutoff.
2022
2023 (I will make another script to wrap up mulitple scripts for
2024 differential calling)
2025
2026 2011-07-07 Tao Liu <[email protected]>
2027 MACS version 2.0.5 (tag:alpha)
2028
2029 * bin/macs2, MACS2/cPeakDetect.py, MACS2/IO/cScoreTrack.pyx,
2030 MACS2/IO/cPeakIO.pyx
2031
2032 Use hash to store peak information. Add back the feature to deal
2033 with data without control.
2034
2035 Fix bug which incorrectly allows small peaks at the end of
2036 chromosomes.
2037
2038 * bin/bdgpeakcall, bin/bdgcmp
2039
2040 Fix bugs. bdgpeakcall can output encodePeak format.
2041
2042 2011-06-22 Tao Liu <[email protected]>
2043 MACS version 2.0.4 (tag:alpha)
2044
2045 * cPeakDetect.py
2046
2047 Fix a bug, correctly assign lambda_bg while --to-small is
2048 set. Thanks Junya Seo!
2049
2050 Add rank and num of bp columns to pvalue-qvalue table.
2051
2052 * cScoreTrack.py
2053
2054 Fix bugs to correctly deal with peakless chromosomes. Thanks
2055 Vaibhav Jain!
2056
2057 Use AFDR for independent tests instead.
2058
2059 * encodePeak
2060
2061 Now MACS can output peak coordinates together with pvalue, qvalue,
2062 summit positions in a single encodePeak format (designed for
2063 ENCODE project) file. This file can be loaded to UCSC
2064 browser. Definition of some specific columns are: 5th:
2065 int(-log10pvalue*10), 7th: fold-change, 8th: -log10pvalue, 9th:
2066 -log10qvalue, 10th: relative summit position to peak start.
2067
2068
2069 2011-06-19 Tao Liu <[email protected]>
2070 MACS version 2.0.3 (tag:alpha)
2071
2072 * Rich output with qvalue, fold enrichment, and pileup height
2073
2074 Calculate q-values using a refined Benjamini–Hochberg–Yekutieli
2075 procedure:
2076
2077 http://en.wikipedia.org/wiki/False_discovery_rate#Dependent_tests
2078
2079 Now we have a similiar xls output file as before. The differences
2080 from previous file are:
2081
2082 1. Summit now is absolute summit, instead of relative summit
2083 position;
2084 2. 'Pileup' is previous 'tag' column. It's the extended fragment
2085 pileup at the peak summit;
2086 3. We now use '-log10(pvalue)' instead of '-10log10(pvalue)', so
2087 5.00 means 1e-5, simple and less confusing.
2088 4. FDR column becomes '-log10(qvalue)' column.
2089 5. The pileup, -log10pvalue, fold_enrichment and -log10qvalue are
2090 the values at the peak summit.
2091
2092 * Extra output files
2093
2094 NAME_pqtable.txt contains pvalue and qvalue relationships.
2095
2096 NAME_treat_pvalue.bdg and NAME_treat_qvalue.bdg store -log10pvalue
2097 and -log10qvalue scores in BedGraph format. Nearby regions with
2098 the same value are not merged.
2099
2100 * Separation of FeatIO.py
2101
2102 Its content has been divided into cPeakIO.pyx, cBedGraph.pyx, and
2103 cFixWidthTrack.pyx. A modified bedGraphTrackI class was
2104 implemented to store pileup, local lambda, pvalue, and qvalue
2105 alltogether in cScoreTrack.pyx.
2106
2107 * Experimental option --half-ext
2108
2109 Suggested by NPS algorithm, I added an experimental option
2110 --half-ext to let MACS only extends ChIP fragment around its
2111 middle point for only 1/2 d.
2112
2113 2011-06-12 Tao Liu <[email protected]>
2114 MACS version 2.0.2 (tag:alpha)
2115
2116 * macs2
2117
2118 Add an error check to see if there is no common chromosome names
2119 from treatment file and control file
2120
2121 * cPeakDetect.pyx, cFeatIO.pyx, cPileup.pyx
2122
2123 Reduce memory usage by removing deepcopy() calls.
2124
2125 * Modify README documents and others.
2126
2127 2011-05-19 Tao Liu <[email protected]>
2128 MACS Version 2.0.1 (tag:alpha)
2129
2130 * cPileup.pyx, cPeakDetect.pyx and peak calling process
2131
2132 Jie suggested me a brilliant simple method to pileup fragments
2133 into bedGraph track. It works extremely faster than the previous
2134 function, i.e, faster than MACS1.3 or MACS1.4. So I can include
2135 large local lambda calculation in MACSv2 now. Now I generate three
2136 bedGraphs for d-size local bias, slocal-size and llocal-size local
2137 bias, and calculate the maximum local bias as local lambda
2138 bedGraph track.
2139
2140 Minor: add_loc in bedGraphTrackI now can correctly merge the
2141 region with its preceding region if their value are the same.
2142
2143 * macs2
2144
2145 Add an option to shift control tags before extension. By default,
2146 control tags will be extended to both sides regardless of strand
2147 information.
2148
2149 2011-05-17 Tao Liu <[email protected]>
2150 MACS Version 2.0.0 (tag:alpha)
2151
2152 * Use bedGraph type to store data internally and externally.
2153
2154 We can have theoretically one-basepair resolution profiles. 10
2155 times smaller in filesize and even smaller after converting to
2156 bigWig for visualization.
2157
2158 * Peak calling process modified. Better peak boundary detection.
2159
2160 Extend ChIP tag to d, and pileup to have a ChIP bedGraph. Extend
2161 Control tag to d and 1,000bp, and pileup to two bedGraphs. (1000bp
2162 one will be averaged to d size) Then calculate the maximum value
2163 of these two tracks and a global background, to have a
2164 local-lambda bedGraph.
2165
2166 Use -10log10poisson_pvalue as scores to generate a score track
2167 before peak calling.
2168
2169 A general peak calling based on a score cutoff, min length of peak
2170 and max gap between nearby peaks.
2171
2172 * Option changes.
2173
2174 Wiggle file output is removed. Now we only support bedGraph
2175 output. The generation of bedGraph is highly recommended since it
2176 will not cost extra time. In other words, bedGraph generation is
2177 internally run even you don't want to save bedGraphs on disk, due
2178 to the peak calling algorithm in MACS v2.
2179
2180 * cProb.pyx
2181
2182 We now can calculate poisson pvalue in log space so that the score
2183 (-10*log10pvalue) will not have a upper limit of 3100 due to
2184 precision of float number.
2185
2186 * Cython is adopted to speed up Python code.
2187
2188 2011-02-28 Tao Liu <[email protected]>
2189 Small fixes
2190
2191 * Replaced with a newest WigTrackI class and fixed the wignorm script.
2192
2193 2011-02-21 Tao Liu <[email protected]>
2194 Version 1.4.0rc2 (Valentine)
2195
2196 * --single-wig option is renamed to --single-profile
2197
2198 * BedGraph output with --bdg or -B option.
2199
2200 The BedGraph output provides 1bp resolution fragment pileup
2201 profile. File size is smaller than wig file. This option can be
2202 combined with --single-profile option to produce a bedgraph file
2203 for the whole genome. This option can also make --space,
2204 --call-subpeaks invalid.
2205
2206 * Fix the description of --shiftsize to correctly state that the
2207 value is 1/2 d (fragment size).
2208
2209 * Fix a bug in the call to __filter_w_control_tags when control is
2210 not available.
2211
2212 * Fix a bug on --to-small option. Now it works as expected.
2213
2214 * Fix a bug while counting the tags in candidate peak region, an
2215 extra tag may be included. (Thanks to Jake Biesinger!)
2216
2217 * Fix the bug for the peaks extended outside of chromosome
2218 start. If the minus strand tag goes outside of chromosome start
2219 after extension of d, it will be thrown out.
2220
2221 * Post-process script for a combined wig file:
2222
2223 The "wignorm" command can be called after a full run of MACS14 as
2224 a postprocess. wignorm can calculate the local background from the
2225 control wig file from MACS14, then use either foldchange,
2226 -10*log10(pvalue) from possion test, or difference after asinh
2227 transformation as the score to build a single wig track to
2228 represent the binding strength. This script will take a
2229 significant long time to process.
2230
2231 * --wigextend has been obsoleted.
2232
2233 2010-09-21 Tao Liu <[email protected]>
2234 Version 1.4.0rc1 (Starry Sky)
2235
2236 * Duplicate reads option
2237
2238 --keep-dup behavior is changed. Now user can specify how many
2239 reads he/she wants to keep at the same genomic location. 'auto' to
2240 let MACS decide the number based on binomial distribution, 'all'
2241 to let MACS keep all reads.
2242
2243 * pvalue and FDR fixes (Thanks to Prof. Zhiping Weng)
2244
2245 By default, MACS will now scale the smaller dataset to the bigger
2246 dataset. For instance, if IP has 10 million reads, and Input has 5
2247 million, MACS will double the lambda value calculated from Input
2248 reads while calling BOTH the positive peaks and negative
2249 peaks. This will address the issue caused by unbalanced numbers of
2250 reads from IP and Input. If --to-small is turned on, MACS will
2251 scale the larger dataset to the smaller one. So from now on, if d
2252 is fixed, then the peaks from a MACS call for A vs B should be
2253 identical to the negative peaks from a B vs A.
2254
2255 2010-09-01 Tao Liu <[email protected]>
2256 Version 1.4.0beta (summer wishes)
2257
2258 * New features
2259
2260 ** Model building
2261
2262 The default behavior in the model building step is slightly
2263 changed. When MACS can't find enough pairs to build model
2264 (implemented in alpha version) or the modeled fragment length is
2265 less than 2 times of tag length (implemented in beta version),
2266 MACS will use 2 times of --shiftsize value as fragment length in
2267 the later analysis. --off-auto can turn off this default behavior.
2268
2269 ** Redundant tag filtering
2270
2271 The IO module is rewritten. The redundant tag filtering process
2272 becomes simpler and works as promise. The maximum allowed number
2273 of tags at the exact same location is calculated from the
2274 sequencing depth and genome size using a binomial distribution,
2275 for both TREAMENT and CONTROL separately. ( previously only
2276 TREATMENT is considered ) The exact same location means the same
2277 coordination and the same strand. Then MACS will only keep at most
2278 this number of tags at the exact same location in the following
2279 analysis. An option --keep-dup can let MACS skip the filtering and
2280 keep all the tags. However this may bring in a lot of sequencing
2281 bias, so you may get many false positive peaks.
2282
2283 ** Single wiggle mode
2284
2285 First thing to mention, this is not the score track that I
2286 described before. By default, MACS generates wiggle files for
2287 fragment pileup for every chromosomes separately. When you use
2288 --single-wig option, MACS will generate a single wiggle file for
2289 all the chromosomes so you will get a wig.gz for TREATMENT and
2290 another wig.gz for CONTROL if available.
2291
2292 ** Sniff -- automatic format detection
2293
2294 Now, by default or "-f AUTO", MACS will decide the input file
2295 format automatically. Technically, it will try to read at most
2296 1000 records for the first 10 non-comment lines. If it succeeds,
2297 the format is decided. I recommend not to use AUTO and specify the
2298 right format for your input files, unless you combine different
2299 formats in a single MACS run.
2300
2301 * Options changes
2302
2303 --single-wig and --keep-dup are added. Check previous section in
2304 ChangeLog for detail.
2305
2306 -f (--format) AUTO is now the default option.
2307
2308 --slocal default: 1000
2309 --llocal default: 10000
2310
2311 * Bug fixed
2312
2313 Setup script will stop the installation if python version is not
2314 python2.6 or python2.7.
2315
2316 Local lambda calculation has been changed back. MACS will check
2317 peak_region, slocal( default 1K) and llocal (default 10K) for the
2318 local bias. The previous 200bps default will cause MACS misses
2319 some peaks where the input bias is very sharp.
2320
2321 sam2bed.py script is corrected.
2322
2323 Relative pos in xls output is fixed.
2324
2325 Parser for ELAND_export is fixed to pass some of the no match
2326 lines. And elandexport2bed.py is fixed too. ( however I can't
2327 guarantee that it works on any eland_export files. )
2328
2329 2010-06-04 Tao Liu <[email protected]>
2330 Version 1.4.0alpha2 (be smarter)
2331
2332 * Options changes
2333
2334 --gsize now provides shortcuts for common genomes, including
2335 human, mouse, C. elegans and fruitfly.
2336
2337 --llocal now will be 5000 bps if there is no input file, so that
2338 local lambda doesn't overkill enriched binding sites.
2339
2340 2010-06-02 Tao Liu <[email protected]>
2341 Version 1.4alpha (be smarter)
2342
2343 * Options changes
2344
2345 --tsize option is redesigned. MACS will use the first 10 lines of
2346 the input to decide the tag size. If user specifies --tsize, it
2347 will override the auto decided tsize.
2348
2349 --lambdaset is replaced by --slocal and --llocal which mean the
2350 small local region and large local region.
2351
2352 --bw has no effect on the scan-window size now. It only affects the
2353 paired-peaks model process.
2354
2355 * Model building
2356
2357 During the model building, MACS will pick out the enriched regions
2358 which are not too high and not too low to build the paired-peak
2359 model. Default the region is from fold 10 to fold 30. If MACS
2360 fails to build the model, by default it will use the nomodel
2361 settings, like shiftsize=100bps, to shift and extend each
2362 tags. This behavior can be turned off by '--off-auto'.
2363
2364 * Output files
2365
2366 An extra file including all the summit positions are saved in
2367 *_summits.bed file. An option '--call-subpeaks' will invoke
2368 PeakSplitter developed by Mali Salmon to split wide peaks into
2369 smaller subpeaks.
2370
2371 * Sniff ( will in beta )
2372
2373 Automatically recognize the input file format, so use can combine
2374 different format in one MACS run.
2375
2376 Not implemented features/TODO:
2377
2378 * Algorithms ( in near future? )
2379
2380 MACS will try to refine the peak boundaries by calculating the
2381 scores for every point in the candidate peak regions. The score
2382 will be the -10*log(10,pvalue) on a local poisson distribution. A
2383 cutoff specified by users (--pvalue) will be applied to find the
2384 precise sub-peaks in the original candidate peak region. Peak
2385 boudaries and peak summits positions will be saved in separate BED
2386 files.
2387
2388 * Single wiggle track ( in near future? )
2389
2390 A single wiggle track will be generated to save the scores within
2391 candidate peak regions in the 10bps resolution. The wiggle file
2392 is in fixedStep format.
2393
2394
2395 2009-10-16 Tao Liu <[email protected]>
2396 Version 1.3.7.1 (Oktoberfest, bug fixed #1)
2397
2398 * bin/Constants.py
2399
2400 Fixed typo. FCSTEP -> FESTEP
2401
2402 * lib/PeakDetect.py
2403
2404 The 'femax' attribute bug is fixed
2405
2406 2009-10-02 Tao Liu <[email protected]>
2407 Version 1.3.7 (Oktoberfest)
2408
2409 * bin/macs, lib/PeakDetect.py, lib/IO/__init__.py, lib/OptValidator.py
2410
2411 Enhancements by Peter Chines:
2412
2413 1. gzip files are supported.
2414 2. when --diag is on, user can set the increment and endpoint for
2415 fold enrichment analysis by setting --fe-step and --fe-max.
2416
2417 Enhancements by Davide Cittaro:
2418
2419 1. BAM and SAM formats are supported.
2420 2. small changes in the header lines of wiggle output.
2421
2422 Enhancements by Me:
2423 1. I added --fe-min option;
2424 2. Bowtie ascii output with suffix ".map" is supported.
2425
2426 Bug fixed:
2427
2428 1. --nolambda bug is fixed. ( reported by Martin in JHU )
2429 2. --diag bug is fixed. ( reported by Bogdan Tanasa )
2430 3. Function to remove suffix '.fa' is fixed. ( reported by Jeff Johnston )
2431 4. Some "fold change" have been changed to "fold enrichment".
2432
2433 2009-06-10 Tao Liu <[email protected]>
2434 Version 1.3.6.1 (default parameter change)
2435
2436 * bin/macs, lib/PeakDetect.py
2437
2438 "--oldfdr" is removed. The 'oldfdr' behaviour becomes
2439 default. "--futurefdr" is added which can turn on the 'new' method
2440 introduced in 1.3.6. By default it's off.
2441
2442 * lib/PeakDetect.py
2443
2444 Fixed a bug. p-value is corrected a little bit.
2445
2446
2447 2009-05-11 Tao Liu <[email protected]>
2448 Version 1.3.6 (Birthday cake)
2449
2450 * bin/macs
2451
2452 "track name" is added to the header of BED output file.
2453
2454 Now the default peak detection method is to consider 5k and 10k
2455 nearby regions in treatment data and peak location, 1k, 5k, and
2456 10k regions in control data to calculate local bias. The old
2457 method can be called through '--old' option.
2458
2459 Information about how many total/unique tags in treatment or
2460 control will be saved in final .xls output.
2461
2462 * lib/IO/__init__.py
2463
2464 ".fa" will be removed from input tag alignment so only the
2465 chromosome names are kept.
2466
2467 WigTrackI class is added for Wiggle like data structure. (not used
2468 now)
2469
2470 The parser for ELAND multi PET files has been fixed. Now the 5'
2471 tag position for a pair will be kept, whereas in the previous
2472 version, the middle points are kept.
2473
2474 * lib/IO/BinKeeper.py
2475
2476 BinKeeperI class is inspired by Jim Kent's library for UCSC genome
2477 browser, which can quickly access certain region for values in a
2478 large wiggle like data file. (not used now)
2479
2480 * lib/OptValidator.py
2481
2482 typo fixed.
2483
2484 * lib/PeakDetect.py
2485
2486 Now the default peak detection method is to consider 5k and 10k
2487 nearby regions in treatment data and peak location, 1k, 5k, and
2488 10k regions in control data to calculate local bias. The old
2489 method can be called through '--old' option.
2490
2491 Two columns have beed added to BED output file. 4th column: peak
2492 name; 5th column: peak score using -10log(10,pvalue) as score.
2493
2494 * setup.py
2495
2496 Add support to build a Mac App through 'setup.py py2app', or a
2497 Windows executable through 'setup.py py2exe'. You need to install
2498 py2app or py2exe package in order to use these functions.
2499
2500 2009-02-12 Tao Liu <[email protected]>
2501 Version 1.3.5 (local lambda fixed, typo fixed, model figure improved)
2502
2503 * PeakDetect.py
2504
2505 Now, besides 1k, 5k, 10k, MACS will also consider peak size region
2506 in control data to calculate local lambda for each peak. Peak
2507 calling results will be slightly different with previous version,
2508 beware!
2509
2510 * OptValidator.py
2511
2512 Typo fixed, ELANDParser -> ELANDResultParser
2513
2514 * OutputWriter.py
2515
2516 Now, modeled d value will be shown on the model figure.
2517
2518 2009-01-06 Tao Liu <[email protected]>
2519 Version 1.3.4 (Happy New Year Version, bug fixed, ELAND multi/PET support)
2520
2521 * macs, IO/__init__.py, PeakDetect.py
2522
2523 Add support for ELAND multi format. Add support for Pair-End
2524 experiment, in this case, 5'end and 3'end ELAND multi format files
2525 are required for treatment or control data. See 00README file for
2526 detail.
2527
2528 Add wigextend option.
2529
2530 Add petdist option for Pair-End Tag experiment, which is the best
2531 distance between 5' and 3' tags.
2532
2533 * PeakDetect.py
2534
2535 Fixed a bug which cause the end positions of every peak region
2536 incorrectly added by 1 bp. ( Thanks Mali Salmon!)
2537
2538 * OutputWriter.py
2539
2540 Fix bugs while generating wiggle files. The start position of
2541 wiggle file is set to 1 instead of 0.
2542
2543 Fix a bug that every 10M bps, signals in the first 'd' range are
2544 lower than actual. ( Thanks Mali Salmon!)
2545
2546
2547 2008-12-03 Tao Liu <[email protected]>
2548 Version 1.3.3 (wiggle bugs fixed)
2549
2550 * OutputWriter.py
2551
2552 Fix bugs while generating wiggle files. 1. 'span=' is added to
2553 'variableStep' line; 2. previously, every 10M bps, the coordinates
2554 were wrongly shifted to the right for 'd' basepairs.
2555
2556 * macs, PeakDetect.py
2557
2558 Add an option to save wiggle files on different resolution.
2559
2560 2008-10-02 Tao Liu <[email protected]>
2561 Version 1.3.2 (tiny bugs fixed)
2562
2563 * IO/__init__.py
2564
2565 Fix 65536 -> 65535. ( Thank Joon)
2566
2567 * Prob.py
2568
2569 Improved for binomial function with extra large number. Imported
2570 from Cistrome project.
2571
2572 * PeakDetect.py
2573
2574 If treatment channel misses reads in some chromosome included in
2575 control channel, or vice versa, MACS will not exit. (Thank Shaun
2576 Mahony)
2577
2578 Instead, MACS will fake a tag at position -1 when calling
2579 treatment peaks vs control, but will ignore the chromosome while
2580 calling negative peaks.
2581
2582 2008-09-04 Tao Liu <[email protected]>
2583 Version 1.3.1 (tiny bugs fixed version)
2584
2585 * Prob.py
2586
2587 Hyunjin Gene Shin contributed some codes to Prob.py. Now the
2588 binomial functions can tolerate large and small numbers.
2589
2590 * IO/__init__.py
2591
2592 Parsers now split lines in BED/ELAND file using any
2593 whitespaces. 'track' or 'browser' lines will be regarded as
2594 comment lines. A bug fixed when throwing StrandFormatError. The
2595 maximum redundant tag number at a single position can be no less
2596 than 65536.
2597
2598
2599 2008-07-15 Tao Liu <[email protected]>
2600 Version 1.3 (naming clarification version)
2601
2602 * Naming clarification changes according to our manuscript:
2603
2604 'frag_len' is changed to 'd'.
2605
2606 'fold_change' is changed to 'fold_enrichment'.
2607
2608 Suggest '--bw' parameter to be determined by users from the real
2609 sonication size.
2610
2611 Maximum FDR is 100% in the output file.
2612
2613 And other clarifications in 00README file and the documents on the
2614 website.
2615
2616 * IO/__init__.py
2617 If the redundant tag number at a single position is over 32767,
2618 just remember 32767, instead of raising an overflow exception.
2619
2620 * setup.py
2621 fixed a typo.
2622
2623 * PeakDetect.py
2624 Bug fixed for diagnosis report.
2625
2626
2627 2008-07-10 Tao Liu <[email protected]>
2628 Version 1.2.2gamma
2629
2630 * Serious bugs fix:
2631
2632 Poisson distribution CDF and inverse CDF functions are
2633 corrected. They can produce right results even for huge lambda
2634 now. So that the p-value and FDR values in the final excel sheet
2635 are corrected.
2636
2637 IO package now can tolerate some rare cases; ELANDParser in IO
2638 package is fixed. (Thank Bogdan)
2639
2640 * Improvement:
2641
2642 Reverse paired peaks in model are rejected. So there will be no
2643 negative 'frag_len'. (Thank Bogdan)
2644
2645 * Features added:
2646
2647 Diagnosis function is completed. Which can output a table file for
2648 users to estimate their sequencing depth.
2649
2650
2651 2008-06-30 Tao Liu <[email protected]>
2652 Version 1.2
2653
2654 * Probe.py is added!
2655
2656 GSL is totally removed from MACS. Instead, I have implemented the
2657 CDF and inverse CDF for poisson and binomial distribution purely
2658 in python.
2659
2660 * Constants.py is added!
2661
2662 Organize constants used in MACS in the Constants.py file.
2663
2664 * All other files are modified!
2665
2666 Foldchange calculation is modified. Now the foldchange only be
2667 calculated at the peak summit position instead of the whole peak
2668 region. The values will be higher and more robust than before.
2669
2670 Features added:
2671
2672 1. MACS can save wiggle format files containing the tag number at
2673 every 10 bp along the genome. Tags are shifted according to our
2674 model before they are calculated.
2675
2676 2. Model building and local lambda calculation can be skipped with
2677 certain options.
2678
2679 3. A diagnosis report can be generated through '--diag'
2680 option. This report can help you get an assumption about the
2681 sequencing saturation. This funtion is only in beta stage.
2682
2683 4. FDR calculation speed is highly improved.
2684
2685 2008-05-28 Tao Liu <[email protected]>
2686 Version 1.1
2687
2688 * TabIO, PeakModel.py ...
2689 Bug fixed to let MACS tolerate some cases while there is no tag on
2690 either plus strand or minus strand.
2691
2692 * setup.py
2693 Check the version of python. If the version is lower than 2.4,
2694 refuse to install with warning.