| 1 |
Multi-series analysis |
| 2 |
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ |
| 3 |
|
| 4 |
.. _multiseries: |
| 5 |
|
| 6 |
Sometimes SAXS data is collected in a way where multiple separate datasets |
| 7 |
need to be processed together, either averaged or subtracted. A simple |
| 8 |
example is if an IEC-SAXS experiment is done and background subtraction is |
| 9 |
done by measuring a blank injection in the same gradient and doing a point |
| 10 |
by point subtraction of the series with sample and the blank series. |
| 11 |
A more complicated example would be how time-resolved SAXS data is collected |
| 12 |
at the BioCAT beamline, where multiple sequential scans along a microfluidic |
| 13 |
mixer are collected for a single injection and some of the scans are used |
| 14 |
for buffer subtraction while others contain scattering from the sample. |
| 15 |
|
| 16 |
The multi-series analysis tool in RAW provides a robust mechanism for loading |
| 17 |
in multiple series at once and carrying out point-by-point averaging and |
| 18 |
subtraction across series. Once a subtracted series is created, you can |
| 19 |
then carry out further data refinement including truncating and binning q ranges, |
| 20 |
averaging together multiple points in a series to improve signal to noise, |
| 21 |
removing particular profiles from the series, and calibrating the series. |
| 22 |
|
| 23 |
The written version of the tutorial follows. |
| 24 |
|
| 25 |
Time-resolved SAXS analysis |
| 26 |
***************************** |
| 27 |
|
| 28 |
Time resolved SAXS data from the BioCAT beamline is used for this tutorial. |
| 29 |
The experiment is a refolding experiment on Cytochrome C. The protein starts |
| 30 |
out chemically denatured in a 4.5 M guanidine buffer. It is then diluted 10x |
| 31 |
into a refolding buffer using a very fast continuous flow microfluidic mixer, |
| 32 |
initiating the refolding reaction. Timepoints after mixing are obtained by |
| 33 |
measuring the scattering some distance along the mixing channel, and are |
| 34 |
determined by how long the solution takes to flow from the mixing point to |
| 35 |
the measurement point. |
| 36 |
|
| 37 |
The basic data collection procedure is as follows. First, flow of mixing and |
| 38 |
sample buffers is started. Then, simultaneously, the X-ray exposure and a |
| 39 |
continuous scan of the mixer is started. The X-ray beam is scanned along the |
| 40 |
observation region, and images are measured while the mixer is moving (a |
| 41 |
continuous/fly scan rather than step scan), which is important for minimizing |
| 42 |
radiation damage and maximizing throughput. Each exposure at a new position |
| 43 |
along the observation region corresponds to a different timepoint after mixing. |
| 44 |
Repeated scans along the mixer add additional data at the same timepoints, which |
| 45 |
improves the signal to noise of the measurement at the timepoint. After |
| 46 |
sufficient buffer scans, the sample is injected into the mixer via an injection |
| 47 |
valve. Measurements are carried out while all the sample flows through the mixer, |
| 48 |
yielding multiple scans with mixed sample measured at every timepoint. Once |
| 49 |
all the sample has passed through the mixer, additional scans of just buffer |
| 50 |
are measured, yielding buffer measurements pre- and post-sample injection and sample |
| 51 |
measurements at every timepoint as part of the same experiment. If necessary, the |
| 52 |
measurement is repeated multiple times to provide good data at each timepoint. |
| 53 |
Typically, at least 3 such measurements are made for a given sample. |
| 54 |
|
| 55 |
|
| 56 |
This tutorial will take you through loading and carrying out buffer subtraction |
| 57 |
on a single measurement, and then averaging multiple such measurements together |
| 58 |
to get a final time series. |
| 59 |
|
| 60 |
#. Clear all of the data in RAW. From the Tools menu select "Multi-Series Analysis" |
| 61 |
to open the multi-series analysis window. |
| 62 |
|
| 63 |
|ms_panel_png| |
| 64 |
|
| 65 |
#. There are three ways that you can load multi-series data in RAW. The primary |
| 66 |
way is to use the Select from disk sub-panel, which allows you to pick a directory |
| 67 |
and then define a filename with variables in it that allows RAW to find the |
| 68 |
series of interest. You can also select series that are loaded in the Series |
| 69 |
control panel in RAW with the "Add from series panel" button and for data |
| 70 |
collected at the BioCAT beamline (or with a compatible file naming convention) |
| 71 |
you can load the data in automatically with the "Auto Select" button. Here |
| 72 |
we will use the Select from disk sub-panel. |
| 73 |
|
| 74 |
|
| 75 |
#. Click the Browse button and select the **Tutorial_Data/multiseres_data/cytc_01** |
| 76 |
folder. This sets the directory where RAW will look for .dat files corresponding |
| 77 |
to your series. |
| 78 |
|
| 79 |
#. The filename field expects two variables, one corresponding to the series |
| 80 |
number, <s>, and one corresponding to the profile number <f>. As with any |
| 81 |
series in RAW, a single series consists of some set of scattering profiles. |
| 82 |
The profile number variable lets you define which profiles (.dat files on disk) |
| 83 |
are in a given series, while the series number defines which series to load. |
| 84 |
Take a look at the data in the **cytc_01** folder. You will see filenames |
| 85 |
such as **cytc_01_005_0001_data_000001_00001.dat** and |
| 86 |
**cytc_01_005_0001_data_000002_00002.dat**. If you scroll down a bit you'll |
| 87 |
see filenames such as **cytc_01_005_0010_data_000001_00001.dat** |
| 88 |
The series number is given by the cytc_01_05\ **_0001** and cytc_01_05\ **_0010** |
| 89 |
values, in this case the first profile is in series 1 (0001) and the second profile |
| 90 |
is in series 10 (0010). The last value in the filename is the profile number, |
| 91 |
so **_00001.dat** is the first profile in the series, **_00002.dat** the second, and |
| 92 |
so on. |
| 93 |
|
| 94 |
* Note: For this time resolved data set, each series represents a single |
| 95 |
scan of the beam down the mixing channel and each file represents a |
| 96 |
single timepoint in that scan. So across all the series the _00001.dat |
| 97 |
profiles are all measuring the first timepoint, the _00002.dat |
| 98 |
profiles the next timepoint, and so on. |
| 99 |
|
| 100 |
* Tip: File naming convention will vary from facility to facility, so |
| 101 |
make sure you understand the convention for your data. |
| 102 |
|
| 103 |
#. In the filename field enter *cytc_01_005_<s>_data_0<f>_<f>.dat*. This is a |
| 104 |
filename constructed to make use of the series number, <s>, and profile number, |
| 105 |
<f> variables that allow RAW to find all the series and files of interest. |
| 106 |
We'll go through this construction piece by piece. |
| 107 |
|
| 108 |
#. If you scroll through all the .dat files in the selected folder, you |
| 109 |
will see that there is an unvarying file prefix: *cytc_01_005_* |
| 110 |
We include this in our filename to match that part of the filename. |
| 111 |
|
| 112 |
#. The *<s>* is put in the series number position. This will let RAW |
| 113 |
match the series numbers, we'll define the exact format of this |
| 114 |
next. |
| 115 |
|
| 116 |
#. The *_data_0* is again fixed across all files in the folder, so |
| 117 |
similar to the prefix we include that to match the filename. |
| 118 |
|
| 119 |
#. The *<f>_<f>.dat* piece makes use of the profile number variable |
| 120 |
twice. Looking at the files in the folder you'll see that every |
| 121 |
time the last number in the filename increments the number right |
| 122 |
before it does as well, e.g. _000001_00001.dat, _000002_00002.dat, |
| 123 |
and so on. The last number is the profile number, but in order to |
| 124 |
match the filename we need to increment both, so we use the <f> |
| 125 |
variable twice. |
| 126 |
|
| 127 |
|ms_filename_png| |
| 128 |
|
| 129 |
#. Look at the data in the selected folder and make a note of the first and |
| 130 |
last series numbers (should be 0001 and 0090). In the Series # line enter |
| 131 |
1 in the first field and 90 in the second field. In the zero pad field |
| 132 |
enter 4. This tells RAW that wherever you put <s> in the filename it |
| 133 |
should substitute numbers in the range from 1 to 90 (so 1, 2, 3, etc up to 90). |
| 134 |
The zero pad value tells RAW what the string format for these numbers is. In |
| 135 |
this case, it will make the numbers in the filename 4 characters long, and |
| 136 |
any extra characters will be zeros. So series 1 is represented as 0001, |
| 137 |
series 10 as 0010. |
| 138 |
|
| 139 |
* Note: If you changed the zero padding to 3, for example, then you |
| 140 |
would get series strings 001 and 010 for series 1 and 10. |
| 141 |
|
| 142 |
#. Look at the data in the selected folder and make a note of the first and |
| 143 |
last file numbers for a single series (should be 00001 and 00040). In the |
| 144 |
Profiles # line enter 1 in the first field and 40 in the second field. |
| 145 |
In the zero pad field enter 5. This tells RAW that wherever you put <f> |
| 146 |
in the filename it should substitute numbers in the range from 1 to 40, |
| 147 |
with the format of those in the filename defined by the zero padding |
| 148 |
as described above. For example, file number 1 is represented as 00001 |
| 149 |
and file number 10 as 00010. |
| 150 |
|
| 151 |
* Note: For both series and file number, you don't have to list |
| 152 |
everything in your target folder. You could, for example, load |
| 153 |
just series 5 to 35 instead of 1 to 90, if you used that as the |
| 154 |
series number range. |
| 155 |
|
| 156 |
|ms_series_file_numbers_png| |
| 157 |
|
| 158 |
#. An important note is that you can use linux \* and ? wildcards in the filename. |
| 159 |
So in this case we could have defined the filename as *cytc_01_005_<s>_*\ \*\ *<f>.dat*. |
| 160 |
This can be useful if your filename can't be matched strictly with the <s> |
| 161 |
and <f> variables. However, searching for files with wildcards in it |
| 162 |
is significantly slower than getting a defined filename, so you should only |
| 163 |
do this if needed. |
| 164 |
|
| 165 |
#. Now that the directory, filename, and series and file numbers are all |
| 166 |
defined, click the 'Select files' button. This will search for files |
| 167 |
matching the values you provided in the selected folder. Series that it |
| 168 |
successfully finds will be displayed in the left panel. Note that in the left |
| 169 |
panel you will first see a number, this series ID increments sequentially and will be |
| 170 |
used to identify the series in the next part of the analysis. After that you |
| 171 |
will see the series name, in this case it should be something like *cytc_01_005_0001*. |
| 172 |
|
| 173 |
* Tip: Scroll down in the left panel to see all 90 loaded series. |
| 174 |
|
| 175 |
|ms_series_load_png| |
| 176 |
|
| 177 |
#. Click on the first series in the left panel. This will display the series |
| 178 |
location and contents in the Series Info sub-panel. Verify that the profiles contained |
| 179 |
in the series are what you expect. In this case, make sure that each profile |
| 180 |
corresponds to series 1 (has _0001 for the series number) and that the profile |
| 181 |
number is sequentially incrementing from 1 to 40 (ends with _00001.dat |
| 182 |
to _00040.dat). Also verify that the data directory is correct. |
| 183 |
|
| 184 |
* Note: The profiles are not yet loaded into RAW. Because that can take some |
| 185 |
time, due to the large number of profiles involved, you have a chance to |
| 186 |
check and make sure you're loading what you want first. |
| 187 |
|
| 188 |
* Tip: In the left panel, you can rearrange the series order by selecting |
| 189 |
one or more series (clicking, shift clicking, or control clicking as in |
| 190 |
the RAW control panels) and then clicking the Move up or Move down buttons. |
| 191 |
You can also remove series you don't want to include in the analysis |
| 192 |
using the Remove button. |
| 193 |
|
| 194 |
* Tip: The number of profiles is provided in the info panel, and is a good |
| 195 |
way to quickly spot check that you loaded what you expected. |
| 196 |
|
| 197 |
* Try: Click on a few other series in the left list and make sure they |
| 198 |
contain the right profiles. |
| 199 |
|
| 200 |
#. Click the "Next" button to load in the series data (may take a little bit) |
| 201 |
and advance to the next part of the analysis, the series buffer and sample |
| 202 |
range window. |
| 203 |
|
| 204 |
|ms_series_info_png| |
| 205 |
|
| 206 |
#. The series buffer and sample range window allows you to define buffer and sample |
| 207 |
series in a manner similar to how you define buffer and sample profiles |
| 208 |
for a LC Series dataset. The window shows a plot of total scattering |
| 209 |
intensity for each series. Here, each point corresponds to the sum of |
| 210 |
the total intensity of every profile in the series (for this example data |
| 211 |
set, each point is the sum of intensity from 40 profiles). The series number |
| 212 |
corresponds to the series ID given in the list of series in the loading |
| 213 |
panel. |
| 214 |
|
| 215 |
* Note that series IDs are simply sequential for the series loaded, so they |
| 216 |
do not necessarily correspond to the series number as defined by <s>. |
| 217 |
For example, if you'd rearranged the list so that the series with number |
| 218 |
0002 was first in the list and 0001 was second, then the first point in |
| 219 |
this intensity plot would have series ID 1 and correspond to the series |
| 220 |
with number 0002. Or if you'd loaded series 5 to 35, the first point |
| 221 |
in the plot would be series 5 but have series ID 1. |
| 222 |
|
| 223 |
|ms_intensity_plot_png| |
| 224 |
|
| 225 |
#. First we will define what series are buffer. Unlike the LC Analysis panel, |
| 226 |
this is not defining a single profile as buffer, rather we are saying that |
| 227 |
every profile in the series is a buffer profile. For this time resolved data, |
| 228 |
recall that we first measure a set of scans (each scan is a single series |
| 229 |
when loaded into RAW) with buffer, then inject sample, measure a set of scans |
| 230 |
with sample, and then measure a final set of scans with just buffer. |
| 231 |
Analogous to a SEC-SAXS elution series, the series where sample is |
| 232 |
being measured have more total scattering than those with buffer, leading |
| 233 |
to the peak in this plot. Use the "Add region" button to add two buffer |
| 234 |
regions, one before the sample measurement and one after, and define the |
| 235 |
start and end points appropriately using either the spin boxes or |
| 236 |
the "Pick" button as you would in the LC Series panel. These |
| 237 |
should be around series 1-30 and 70-90. |
| 238 |
|
| 239 |
* Note: Because of the injection method, there is some tailing in the |
| 240 |
sample concentration across later series. So to be conservative we |
| 241 |
pick a smaller post-buffer region near the end of the collected series. |
| 242 |
|
| 243 |
* Tip: To use the interactive range picker, click the Pick button, then |
| 244 |
click on the start point of the range on the plot and then the end |
| 245 |
point of the range on the plot. |
| 246 |
|
| 247 |
* Note: Buffer regions are shown on the plot in green. |
| 248 |
|
| 249 |
* Note: Defining a buffer region is optional. If data is already subtracted |
| 250 |
you can proceed to the next step in the tutorial without doing this. |
| 251 |
|
| 252 |
|ms_buffer_range_png| |
| 253 |
|
| 254 |
#. Next we define what series are sample. Again, this is telling RAW that |
| 255 |
every profile in that series is a sample profile. Use the "Add region" button |
| 256 |
to add a sample region where the peak in the data is. This should be |
| 257 |
around series 32-44. |
| 258 |
|
| 259 |
* Note: Sample regions are shown on the plot in purple. |
| 260 |
|
| 261 |
|ms_sample_range_png| |
| 262 |
|
| 263 |
#. Click the "Next" button to move to the subtracted profiles window and carry out |
| 264 |
the buffer subtraction and averaging. In this step, RAW carries out a |
| 265 |
point-by-point average of the defined buffer region(s) and sample |
| 266 |
region(s). For this data, there are 40 timepoints (one per distinct |
| 267 |
profile number from the loading page) and each timepoint has 90 |
| 268 |
profiles associated with it (one per series number from the loading page) |
| 269 |
So RAW averages the first profile across series 1-30 and 70-90 |
| 270 |
to create the average buffer profile at that timepoint. It then averages the |
| 271 |
first profile across series 32-44 to create the average sample profile at |
| 272 |
that timepoint. The average buffer profile is then subtracted from the average |
| 273 |
sample profile to yield the subtracted scattering profile at that timepoint. |
| 274 |
This is carried out for all timepoints, resulting in 40 subtracted scattering |
| 275 |
profiles, each at a different measured timepoint. RAW then carries out |
| 276 |
automated |Rg| and MW calculations for each subtracted profile and displays |
| 277 |
the results in newly displayed subtracted profiles window. |
| 278 |
|
| 279 |
|ms_profile_panel_png| |
| 280 |
|
| 281 |
#. A single time resolved measurement doesn't always yield great signal to noise |
| 282 |
data. You can see from the subtracted profile plotted at the bottom of the window |
| 283 |
that the data is highly noisy, and thus the automated |Rg| and MW calculations |
| 284 |
were not run successfully. We need to average together multiple measurements |
| 285 |
to improve this signal to noise and yield usable data. |
| 286 |
|
| 287 |
* Note: This dataset represents a worst case scenario. Cytochrome C is |
| 288 |
small, so has lower scattering, and the final renaturing buffer |
| 289 |
contains 0.45 M guanidine, which reduces the contrast. For most time |
| 290 |
resolved datasets a single measurement will yield better signal to noise |
| 291 |
data. |
| 292 |
|
| 293 |
#. We will now apply a time calibration to the data. This requires two things. |
| 294 |
First, the profile header for each subtracted profile must contain a value |
| 295 |
that can be used to calibrate the data. Second, a calibration file that converts |
| 296 |
this header value into the desired value must be provided. A linear interpolation |
| 297 |
of the data in the calibration file is created, f(x) and the calibrated value |
| 298 |
is calculated as cal_val = f(input_val). Click the "Load Calibration" button |
| 299 |
and select the **Tutorial_Data/multiseries_data/time_8_ml_min.csv** file. |
| 300 |
This file contains distance from the mixing region in mm and time after |
| 301 |
mixing in ms. |
| 302 |
|
| 303 |
* Note: A calibration file should consist of two columns in csv format. |
| 304 |
The first column is the input value and the second the output value. |
| 305 |
Any header lines should have # as the first character (e.g. "#distance,ms" |
| 306 |
for the calibration file used here) |
| 307 |
|
| 308 |
#. The "Cal. input key" selection menu lists all the available keys in the profile |
| 309 |
header. For this time resolved data, the calibration is done by the position |
| 310 |
of the beam on the microfluidic mixer. Select the "x" header key in the menu, |
| 311 |
which gives the x position of the mixer. |
| 312 |
|
| 313 |
#. The "Cal. output key" field is the name that the calibrated point will be |
| 314 |
saved with in the profile header. Enter "time". |
| 315 |
|
| 316 |
#. The "Cal offset" field allows you to provide an offset value that will be |
| 317 |
added to the value from the profile header (i.e. cal_val = f(input_val+offset)). |
| 318 |
For this dataset, this is the measured position of the mixing region in x. |
| 319 |
Enter 71.13 for this value. |
| 320 |
|
| 321 |
|ms_calibration_png| |
| 322 |
|
| 323 |
#. Click "Process data" to apply the calibration. After it finishes you'll see |
| 324 |
that the x axis of the plots is now in your calibrated units, in this case |
| 325 |
time in ms. |
| 326 |
|
| 327 |
|ms_calibration2_png| |
| 328 |
|
| 329 |
#. We will now save the settings used to process this data. This allows us to |
| 330 |
apply the same settings to additional datasets, making processing |
| 331 |
quicker. It also provides a record of what settings were used to produce |
| 332 |
the final subtracted series. Click the "Save analysis settings" button |
| 333 |
and save the settings as "cytc_01.json" in the top level |
| 334 |
**Tutorial_Data/multiseries_data** folder. |
| 335 |
|
| 336 |
|ms_save_settings_png| |
| 337 |
|
| 338 |
#. Click the done button at the bottom of the window. This will close the |
| 339 |
analysis window and send the subtracted series to the Series control panel |
| 340 |
and Series plot, where you can do further analysis on it. |
| 341 |
|
| 342 |
#. Open the Multi-Series Analysis window. Click the "Load analysis settings" |
| 343 |
button to load in the **cytc_01.json** settings you just saved. You will see |
| 344 |
the "Select from disk" controls fill in the previous values, and additionally |
| 345 |
the range settings and settings from the subsequent windows are also loaded. |
| 346 |
|
| 347 |
|ms_load_settings_png| |
| 348 |
|
| 349 |
#. Since this data is from BioCAT, it is compatible with the Auto select loading |
| 350 |
function. Click the "Auto select" button and select any profile in the |
| 351 |
**Tutorial_Data/mutliseries_data/cytc_03** folder. You will see the series |
| 352 |
load in the left panel as before. |
| 353 |
|
| 354 |
|ms_auto_load_png| |
| 355 |
|
| 356 |
#. Click the "Next" button to move to the buffer and sample range selection |
| 357 |
window. Note that the previously used ranges are selected by default. |
| 358 |
|
| 359 |
#. The experimental strategy changed slightly for this dataset, so we need to |
| 360 |
tweak the sample and buffer ranges. Change those as appropriate. |
| 361 |
|
| 362 |
* Note: Buffer ranges should be around 1-30 and 75-90, and sample range |
| 363 |
around 32-49. |
| 364 |
|
| 365 |
|ms_ranges_repeat_png| |
| 366 |
|
| 367 |
#. Click the "Next" button to move to the subtracted profiles window. Note that |
| 368 |
the time calibration has been applied automatically, as it was loaded with |
| 369 |
the rest of the settings. |
| 370 |
|
| 371 |
#. Save the settings for this analysis as "cytc_03" and click "Done" to send |
| 372 |
the subtracted time series to the Series plot. |
| 373 |
|
| 374 |
#. In the Series Control Panel, save the cytc_01 and cytc_03 series that you |
| 375 |
have processed to the **Tutorial_Data/multiseries_data** folder. |
| 376 |
|
| 377 |
#. Next we will average the time resolved data across all 5 repeated time |
| 378 |
series measurements. This will improve the signal to noise of the final dataset. |
| 379 |
Open the **cytc_02_006_00.hdf5**, **cytc_04_004_0.hdf5** and **cytc_06_018_0.hdf5** |
| 380 |
series in the **Tutorial_Data/multiseries_data** folder. |
| 381 |
|
| 382 |
#. Select all the cytc series in the Series Control Panel, right click on a |
| 383 |
selected series, and select "Multi-Series analysis" from the pop-up menu |
| 384 |
to open a multi-series analysis window with the selected series already |
| 385 |
loaded. |
| 386 |
|
| 387 |
#. In the multi-series analysis window, click "Next" to move to the series |
| 388 |
buffer and sample range window. Add a sample range that spans all 5 series. |
| 389 |
|
| 390 |
|ms_all_sample_range_png| |
| 391 |
|
| 392 |
#. Click "Next" to advance to the subtracted profiles window. |
| 393 |
|
| 394 |
* Note: Because this is an average of 5 repeated measurements, the signal |
| 395 |
to noise is significantly improved and the automated methods are able |
| 396 |
to calculate |Rg| and MW at most timepoints. |
| 397 |
|
| 398 |
#. Because we've previously calibrated each individual series in time, we can |
| 399 |
apply that calibration to the average of all these independent series. Check the |
| 400 |
"Calibration in header" box and from the drop-down menu select "time" |
| 401 |
as the Cal. input key. This tells RAW that the calibration already exists |
| 402 |
in the header, rather than having to be calculated from a header value, |
| 403 |
and that the appropriate key to use for the calibration is "time". |
| 404 |
|
| 405 |
|ms_cal_in_header_png| |
| 406 |
|
| 407 |
#. Click the "Process data" button to reprocess the data with this calibration. |
| 408 |
|
| 409 |
* Note: The plots will now show the x axis with calibrated time values. |
| 410 |
For this dataset the time unit is milliseconds, so all the data is |
| 411 |
recorded for times less than 1 ms after mixing. |
| 412 |
|
| 413 |
|ms_average_data_png| |
| 414 |
|
| 415 |
#. Next we will bin and trim the data to improve the data quality of the individual |
| 416 |
scattering profiles at each timepoint. We'll start with q binning. Check |
| 417 |
the "Rebin q" box, set the Q bin type to "Linear", set the Q bin mode to |
| 418 |
"Points" and set the number of Q bin points to 150. |
| 419 |
|
| 420 |
* Note: Q bin type can be linear or logarithmic, and dictates whether the |
| 421 |
bin spacing is linear or log base 10 in q. Q bin mode can be factor, which |
| 422 |
reduces the number of q points in the profile by the provided factor, |
| 423 |
or number of points, which sets the number of q points in the profile |
| 424 |
to a fixed value. |
| 425 |
|
| 426 |
|ms_q_bin_png| |
| 427 |
|
| 428 |
#. Click the "Process data" button to reprocess the data with this q binning. |
| 429 |
The plotted profile should now show significantly less noise. |
| 430 |
|
| 431 |
|ms_q_bin_profile_png| |
| 432 |
|
| 433 |
#. Because of strong parasitic scattering from the mixer window at some positions, |
| 434 |
the usable q range does not necessarily extend to the lowest q point for this |
| 435 |
dataset, as seen from the dip below zero in the intensity at the lowest q |
| 436 |
point in the plotted profile. Right click on the profile plot and select |
| 437 |
"Log-Log" to change to a logarithmic x axis to see the low q better. |
| 438 |
Then use the "Plot profile" selector to scroll through the profiles at |
| 439 |
different timepoints and get a feel for where the first usable q point is. |
| 440 |
|
| 441 |
* Note: The vertical line on the plot shows which |Rg|/I(0)/MW values correspond |
| 442 |
to the displayed profile. The value in the Plot profile selector box |
| 443 |
is the x axis value corresponding to the plotted profile, in this case |
| 444 |
the time in ms. |
| 445 |
|
| 446 |
* Tip: By right clicking on the parameter plots (e.g. |Rg|) you can change |
| 447 |
the plot type to be linear or logarithmic on either y or x axes. |
| 448 |
|
| 449 |
|ms_plot_profile_png| |
| 450 |
|
| 451 |
#. As you scroll through the profiles you should notice significant variation |
| 452 |
at low q below ~0.01 1/Å. In the first box of the "Q range" line, type 0.01 |
| 453 |
and then the enter key. This control will automatically pick the nearest q value to |
| 454 |
the value you type. |
| 455 |
|
| 456 |
|ms_q_min_png| |
| 457 |
|
| 458 |
#. Click the "Process data" button to set this q min value and then scroll back |
| 459 |
through the profiles to see if you need to adjust the q min value further. |
| 460 |
|
| 461 |
#. You should see there's still more variation than would be expected, indicating |
| 462 |
a higher noise level than desired at the lowest q points. Se the q min value |
| 463 |
to 0.015, and then process the data again. |
| 464 |
|
| 465 |
#. Put the profiles plot back on Log-lin axes and scroll through the profiles, |
| 466 |
looking now at the signal in the high q range. Again because these are relatively |
| 467 |
low signal to noise experiments, the usable q range may not extend all the way |
| 468 |
out to the maximum measured q point, so look for the q range where about |
| 469 |
half the data points are negative, indicating the sample signal is not |
| 470 |
significantly above the background. |
| 471 |
|
| 472 |
Tip: If unsubtracted sample and buffer profiles match, the no signal |
| 473 |
condition, then just from noise you'd expect half the points in the |
| 474 |
subtracted profile to be a bit below zero and half a bit above zero. |
| 475 |
|
| 476 |
#. Set the maximum q value to the highest usable q value for the profiles |
| 477 |
that you found, which should be around 0.5-0.55 1/Å and then process the |
| 478 |
data again. |
| 479 |
|
| 480 |
* Note: The usable q range for these time resolved SAXS experiments will |
| 481 |
depend on the sample size and concentration and buffer composition. |
| 482 |
Because of the small size (12.3 kDa) and high denaturant concentration |
| 483 |
these cytochrome c refolding experiments are a worst case for signal |
| 484 |
to noise. Often a wider q range is accessible, though due to the strong |
| 485 |
parasitic scattering from the mixer it is rare to get a usable low q |
| 486 |
below ~0.01 1/Å. |
| 487 |
|
| 488 |
|ms_q_range_png| |
| 489 |
|
| 490 |
|
| 491 |
#. The series panel can also be used to remove outlier points from the series. |
| 492 |
For example, sometimes there's a bit of dust on the mixer window that leads |
| 493 |
to very strong and extended in q background scattering at a particular timepoint, |
| 494 |
rendering that timepoint not not usable, or perhaps the first timepoint |
| 495 |
starts too close to the edge of the mixer and there is significant additional |
| 496 |
scattering due to that. For this dataset we will excluding only the earliest |
| 497 |
timepoint, as that has excess scattering from the edge of the mixer. |
| 498 |
|
| 499 |
* Note: You can't always tell this from a casual examination of the profiles. |
| 500 |
Looking at the images collected at each timepoint can help here. |
| 501 |
|
| 502 |
|
| 503 |
#. Mouse over the first data point on the |Rg| plot. At the bottom of the plot |
| 504 |
you will see the plot values corresponding to your mouse position. Make |
| 505 |
a note of the frame number (should be 0). |
| 506 |
|
| 507 |
* Tip: To make sure you're on the right point it can help to change the |
| 508 |
plotted profile to the first one, but this isn't necessary. |
| 509 |
|
| 510 |
|ms_plot_values_png| |
| 511 |
|
| 512 |
#. In the Exclude profiles list enter the frame number of the profile to |
| 513 |
exclude from the series, in this case "0". |
| 514 |
|
| 515 |
* Note: If you want to exclude multiple profiles, you would enter those |
| 516 |
as a comma list. For example, putting "0,5,15,30" in the |
| 517 |
excluded profiles field would exclude frames 0, 5, 15, and 30. |
| 518 |
|
| 519 |
|ms_exclude_profiles_png| |
| 520 |
|
| 521 |
#. Process the data with the new settings. Verify on the plot that the first |
| 522 |
frame used is now 1. |
| 523 |
|
| 524 |
#. If the data seems to be significantly oversampling the change of interest |
| 525 |
you can bin together timepoints to improve the signal to noise. In this case, |
| 526 |
we see that there's a gradual change from an |Rg| of ~23-24 Å to an |Rg| |
| 527 |
of ~18-19 Å over the measured timepoints, but there is no fast change |
| 528 |
observed so we can do this binning. |
| 529 |
|
| 530 |
#. Check the "Rebin series" box and set the "Series bin factor" to 2. This will |
| 531 |
average together every two adjacent profiles. Process the data to apply this |
| 532 |
binning. You should now see half the number of timepoints in the parameters |
| 533 |
plot. |
| 534 |
|
| 535 |
* Note: The resulting calibration value is the average calibration value |
| 536 |
across the averaged profiles. The frame number is reported as the first frame |
| 537 |
number of the averaged profiles, e.g. an average of profiles 1 and 2 will |
| 538 |
report frame 1 as the frame value. |
| 539 |
|
| 540 |
* Tip: If you want to preserve the individual profile sampling but smooth |
| 541 |
out the calculated |Rg|, I(0), and MW values in a similar fashion you can |
| 542 |
change the "Series average window" value. This applies a rolling average |
| 543 |
in the same way that these values are calculated for an LC Series, |
| 544 |
so setting this value to 2, for example, would calculate the values using |
| 545 |
a rolling average window of size 2 across the series. |
| 546 |
|
| 547 |
|ms_series_bin_png| |
| 548 |
|
| 549 |
#. The series is now fully processed. Save your analysis settings in the |
| 550 |
**Tutorial_Data/multiseries_data** folder as cytc_avg.json. This will |
| 551 |
allow you to reload the settings and tweak your processing if desired. |
| 552 |
|
| 553 |
#. To make for easy analysis or plotting outside of RAW, you can export the |
| 554 |
parameter plot data to csv. Right click on the |Rg| plot (or the I(0) or MW |
| 555 |
plots) and select "Export data as CVS", then save it in the |
| 556 |
**Tutorial_data/multiseries_data** folder as cytc_avg_timeseries.csv. |
| 557 |
|
| 558 |
* Note: All this data is available from the individual profiles, and |
| 559 |
|Rg|, I(0), and MW can be exported from the series control panel. |
| 560 |
This is just a convenient way to get it. |
| 561 |
|
| 562 |
* Try: Open the CSV file in Excel or a similar program. |
| 563 |
|
| 564 |
* Tip: If you fail to save this here, you can reopen the multiseries |
| 565 |
analysis window for just the single processed series, set the sample |
| 566 |
range as just the single series, set the calibration from header, |
| 567 |
and recreate the plots. Note that because this uses the final processed |
| 568 |
series you won't be able to under things like q binning or series binning, |
| 569 |
that would require reanalyzing the original input series. But it's a |
| 570 |
quick way to regenerate these plots and export the data if you need to. |
| 571 |
|
| 572 |
#. The multi-series window also has an additional data exploration capability. |
| 573 |
Check the "Plot multiple profiles" box. The profiles plot will now show |
| 574 |
multiple profiles, based on your settings. The first profile shown is |
| 575 |
the one selected by the Plot profile box. From that, additional profiles |
| 576 |
are plotted using the "Plot every" setting and the "Last profile" selection. |
| 577 |
In this case, every 5th profile after the first selected profile is plotted, |
| 578 |
ending with the last profile. |
| 579 |
|
| 580 |
* Note: The additional vertical lines that appear on the parameters plot |
| 581 |
are color coded to match the plotted profiles in the profiles plot. |
| 582 |
|
| 583 |
|ms_plot_multiple_png| |
| 584 |
|
| 585 |
#. Try adjusting the starting profile, plot every, and last profile settings |
| 586 |
to explore the data a bit. You might try changing the profiles plot type |
| 587 |
as well, such as to a Kratky or Dimensionless Kratky. |
| 588 |
|
| 589 |
#. Click the Done button to finish the analysis and send the series to the |
| 590 |
series tab. Save this processed series. From here you can carry out further |
| 591 |
analysis. For example, you might send individual profiles to the profiles plot to |
| 592 |
analyze individually or using one of the deconvolution tools (EFA or REGALS) |
| 593 |
to extract out concentration vs. time and profiles for individual components |
| 594 |
in solution. For this particular dataset, we were most interested in whether |
| 595 |
we could capture the burst phase that happens between fully denatured (|Rg| ~31 Å) |
| 596 |
and the next intermediate state (|Rg| ~24 Å). Unfortunately, the initial |
| 597 |
rapid collapse seems to be faster than our earliest timepoint (45 microseconds), |
| 598 |
and so this data doesn't provide any additional insight over previously |
| 599 |
published SAXS refolding studies of cytochrome C (which had an earliest |
| 600 |
timepoint of ~150 microseconds). |
| 601 |
|
| 602 |
|
| 603 |
|
| 604 |
.. |Rg| replace:: R\ :sub:`g` |
| 605 |
|
| 606 |
|
| 607 |
.. |ms_panel_png| image:: images/ms_panel.png |
| 608 |
:target: ../_images/ms_panel.png |
| 609 |
|
| 610 |
.. |ms_filename_png| image:: images/ms_filename.png |
| 611 |
:width: 500 px |
| 612 |
:target: ../_images/ms_filename.png |
| 613 |
|
| 614 |
.. |ms_series_file_numbers_png| image:: images/ms_series_file_numbers.png |
| 615 |
:width: 500 px |
| 616 |
:target: ../_images/ms_series_file_numbers.png |
| 617 |
|
| 618 |
.. |ms_series_load_png| image:: images/ms_series_load.png |
| 619 |
:target: ../_images/ms_series_load.png |
| 620 |
|
| 621 |
.. |ms_series_info_png| image:: images/ms_series_info.png |
| 622 |
:target: ../_images/ms_series_info.png |
| 623 |
|
| 624 |
.. |ms_intensity_plot_png| image:: images/ms_intensity_plot.png |
| 625 |
:target: ../_images/ms_intensity_plot.png |
| 626 |
|
| 627 |
.. |ms_buffer_range_png| image:: images/ms_buffer_range.png |
| 628 |
:target: ../_images/ms_buffer_range.png |
| 629 |
|
| 630 |
.. |ms_sample_range_png| image:: images/ms_sample_range.png |
| 631 |
:target: ../_images/ms_sample_range.png |
| 632 |
|
| 633 |
.. |ms_profile_panel_png| image:: images/ms_profile_panel.png |
| 634 |
:target: ../_images/ms_profile_panel.png |
| 635 |
|
| 636 |
.. |ms_calibration_png| image:: images/ms_calibration.png |
| 637 |
:width: 300 px |
| 638 |
:target: ../_images/ms_calibration.png |
| 639 |
|
| 640 |
.. |ms_calibration2_png| image:: images/ms_calibration2.png |
| 641 |
:target: ../_images/ms_calibration2.png |
| 642 |
|
| 643 |
.. |ms_save_settings_png| image:: images/ms_save_settings.png |
| 644 |
:width: 300 px |
| 645 |
:target: ../_images/ms_save_settings.png |
| 646 |
|
| 647 |
.. |ms_load_settings_png| image:: images/ms_load_settings.png |
| 648 |
:target: ../_images/ms_load_settings.png |
| 649 |
:width: 600 px |
| 650 |
|
| 651 |
.. |ms_auto_load_png| image:: images/ms_auto_load.png |
| 652 |
:target: ../_images/ms_auto_load.png |
| 653 |
|
| 654 |
.. |ms_ranges_repeat_png| image:: images/ms_ranges_repeat.png |
| 655 |
:target: ../_images/ms_ranges_repeat.png |
| 656 |
|
| 657 |
.. |ms_all_sample_range_png| image:: images/ms_all_sample_range.png |
| 658 |
:target: ../_images/ms_all_sample_range.png |
| 659 |
|
| 660 |
.. |ms_cal_in_header_png| image:: images/ms_cal_in_header.png |
| 661 |
:width: 300 px |
| 662 |
:target: ../_images/ms_cal_in_header.png |
| 663 |
|
| 664 |
.. |ms_average_data_png| image:: images/ms_average_data.png |
| 665 |
:target: ../_images/ms_average_data.png |
| 666 |
|
| 667 |
.. |ms_q_bin_png| image:: images/ms_q_bin.png |
| 668 |
:width: 300 px |
| 669 |
:target: ../_images/ms_q_bin.png |
| 670 |
|
| 671 |
.. |ms_q_bin_profile_png| image:: images/ms_q_bin_profile.png |
| 672 |
:width: 500 px |
| 673 |
:target: ../_images/ms_q_bin_profile.png |
| 674 |
|
| 675 |
.. |ms_plot_profile_png| image:: images/ms_plot_profile.png |
| 676 |
:target: ../_images/ms_plot_profile.png |
| 677 |
|
| 678 |
.. |ms_q_min_png| image:: images/ms_q_min.png |
| 679 |
:width: 300 px |
| 680 |
:target: ../_images/ms_q_min.png |
| 681 |
|
| 682 |
.. |ms_q_range_png| image:: images/ms_q_range.png |
| 683 |
:target: ../_images/ms_q_range.png |
| 684 |
|
| 685 |
.. |ms_plot_values_png| image:: images/ms_plot_values.png |
| 686 |
:target: ../_images/ms_plot_values.png |
| 687 |
|
| 688 |
.. |ms_exclude_profiles_png| image:: images/ms_exclude_profiles.png |
| 689 |
:width: 300 px |
| 690 |
:target: ../_images/ms_exclude_profiles.png |
| 691 |
|
| 692 |
.. |ms_series_bin_png| image:: images/ms_series_bin.png |
| 693 |
:target: ../_images/ms_series_bin.png |
| 694 |
|
| 695 |
.. |ms_plot_multiple_png| image:: images/ms_plot_multiple.png |
| 696 |
:target: ../_images/ms_plot_multiple.png |