| 1 |
/* |
| 2 |
* MIT License |
| 3 |
* |
| 4 |
* Copyright 2017 Broad Institute |
| 5 |
* |
| 6 |
* Permission is hereby granted, free of charge, to any person obtaining a copy |
| 7 |
* of this software and associated documentation files (the "Software"), to deal |
| 8 |
* in the Software without restriction, including without limitation the rights |
| 9 |
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell |
| 10 |
* copies of the Software, and to permit persons to whom the Software is |
| 11 |
* furnished to do so, subject to the following conditions: |
| 12 |
* |
| 13 |
* The above copyright notice and this permission notice shall be included in all |
| 14 |
* copies or substantial portions of the Software. |
| 15 |
* |
| 16 |
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
| 17 |
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
| 18 |
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE |
| 19 |
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER |
| 20 |
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, |
| 21 |
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE |
| 22 |
* SOFTWARE. |
| 23 |
*/ |
| 24 |
package org.broadinstitute.dropseqrna.barnyard.digitalallelecounts.sampleassignment.multisample; |
| 25 |
|
| 26 |
import java.util.ArrayList; |
| 27 |
import java.util.List; |
| 28 |
import java.util.stream.Collectors; |
| 29 |
|
| 30 |
import org.broadinstitute.dropseqrna.barnyard.digitalallelecounts.LikelihoodUtils; |
| 31 |
|
| 32 |
public class AllPairedSampleAssignmentsForCell { |
| 33 |
|
| 34 |
private SamplePairAssignmentForCell best; |
| 35 |
private List<SamplePairAssignmentForCell> all; |
| 36 |
private final String cell; |
| 37 |
private boolean dataAdded=false; |
| 38 |
private double bestPairPvalue; |
| 39 |
private final boolean scaleLikelihoods; |
| 40 |
|
| 41 |
public AllPairedSampleAssignmentsForCell (final String cell, final boolean scaleLikelihoods) { |
| 42 |
this.cell=cell; |
| 43 |
this.scaleLikelihoods=scaleLikelihoods; |
| 44 |
all=new ArrayList<>(); |
| 45 |
} |
| 46 |
|
| 47 |
public void add (final SamplePairAssignmentForCell s) { |
| 48 |
all.add(s); |
| 49 |
dataAdded=true; |
| 50 |
} |
| 51 |
|
| 52 |
/** |
| 53 |
* This looks for the pair of donors that is most likely to have generated the data |
| 54 |
* It searches through all donor pairs where the results are well defined - they have at least one informative UMI to differentiate the two |
| 55 |
* In the case where no donor pairs have at least one informative UMI (genetically similar donors with very few observations), |
| 56 |
* select the pair with the most data, and set the mixture to be 0.5. It's impossible to say what the correct mixture is, as all 3 |
| 57 |
* results are equally valid. |
| 58 |
*/ |
| 59 |
private void calculateBest() { |
| 60 |
// set the scaling on the likelihoods of all donor pairs before recalculating. |
| 61 |
if (scaleLikelihoods) setScaling(); |
| 62 |
|
| 63 |
// recalculate |
| 64 |
// TODO: should I instead compare all doublet likelihoods when computing the confident model doublet probability? |
| 65 |
// if the number of doublet likelihoods < 2, then the pvalue should be 1. |
| 66 |
// double [] doubleLikelihoods = all.stream().filter(x -> x.getDoubletPvalue()>=0.9d).mapToDouble(x -> x.getScaledDoubletLikelihood()).toArray(); |
| 67 |
double [] likelihoods = all.stream().mapToDouble(x -> x.getScaledDoubletLikelihood()).toArray(); |
| 68 |
double bestPairPvalue = LikelihoodUtils.getInstance().getPvalueFromLog10Likelihood(likelihoods); |
| 69 |
|
| 70 |
double maxLike=-Double.MAX_VALUE; |
| 71 |
SamplePairAssignmentForCell best = null; |
| 72 |
|
| 73 |
for (SamplePairAssignmentForCell spac: all) { |
| 74 |
double bestLike = spac.getScaledBestLikelihood(); |
| 75 |
// important - you need to have informative UMIs for this to be a good call |
| 76 |
if (bestLike>maxLike & spac.getNumInformativeUMIs()>0) { |
| 77 |
maxLike=bestLike; |
| 78 |
best=spac; |
| 79 |
} |
| 80 |
} |
| 81 |
|
| 82 |
this.best=best; |
| 83 |
this.bestPairPvalue=bestPairPvalue; |
| 84 |
dataAdded=false; |
| 85 |
|
| 86 |
// fall back on finding the best uninformative pair. |
| 87 |
if (best==null) calculateBestUninformative(); |
| 88 |
} |
| 89 |
|
| 90 |
/** |
| 91 |
* Sets a scaling metric where donor pairs with fewer UMIs than the max have an additional penalty to their likelihoods to make them more consistent. |
| 92 |
*/ |
| 93 |
private void setScaling () { |
| 94 |
int maxUMIs = all.stream().mapToInt(x -> x.getNumUMIs()).max().getAsInt(); |
| 95 |
for (SamplePairAssignmentForCell s: all) { |
| 96 |
double scaling = (double) maxUMIs / s.getNumUMIs(); |
| 97 |
s.setScalingFactor(scaling); |
| 98 |
} |
| 99 |
} |
| 100 |
|
| 101 |
/** |
| 102 |
* Scale likelihoods |
| 103 |
*/ |
| 104 |
|
| 105 |
/** |
| 106 |
* In the case where no donor pairs have at least one informative UMI (genetically similar donors with very few observations), |
| 107 |
* select the pair with the most data, and set the mixture to be 0.5. It's impossible to say what the correct mixture is, as all 3 |
| 108 |
* results are equally valid. |
| 109 |
* This can also result in the "best" pair having no UMIs at all. |
| 110 |
*/ |
| 111 |
private void calculateBestUninformative() { |
| 112 |
double maxSize=0; |
| 113 |
SamplePairAssignmentForCell best = null; |
| 114 |
for (SamplePairAssignmentForCell spac: all) { |
| 115 |
int size = spac.getNumUMIs(); |
| 116 |
if (size>=maxSize) { |
| 117 |
maxSize=size; |
| 118 |
best=spac; |
| 119 |
} |
| 120 |
} |
| 121 |
|
| 122 |
// construct a new object and set the mixture to be 0.5 |
| 123 |
SamplePairAssignmentForCell result = new SamplePairAssignmentForCell(best.getCellBarcode(), best.getSampleOne(), best.getSampleTwo(), |
| 124 |
best.getSampleOneSingleLikelihood(), best.getSampleTwoSingleLikelihood(), best.getDoubletLikelihood(), -1.0d, best.getImpossibleAllelesSampleOne(), |
| 125 |
best.getImpossibleAllelesSampleTwo(), best.getNumInformativeSNPs(), best.getNumSNPs(), best.getNumUMIs(), best.getNumInformativeUMIs(), |
| 126 |
best.getNumInformativeHomozygousUMIsSampleOne(), best.getNumInformativeHomozygousUMIsSampleTwo()); |
| 127 |
this.best=result; |
| 128 |
|
| 129 |
} |
| 130 |
|
| 131 |
|
| 132 |
|
| 133 |
public String getCell() { |
| 134 |
return cell; |
| 135 |
} |
| 136 |
|
| 137 |
public SamplePairAssignmentForCell getBestAssignment () { |
| 138 |
// if data has been added, recalculate best. |
| 139 |
if (this.dataAdded) calculateBest(); |
| 140 |
return this.best; |
| 141 |
} |
| 142 |
|
| 143 |
public double getBestPairPvalue () { |
| 144 |
// if data has been added, recalculate best. |
| 145 |
if (this.dataAdded) calculateBest(); |
| 146 |
return this.bestPairPvalue; |
| 147 |
} |
| 148 |
|
| 149 |
public List<SamplePairAssignmentForCell> getOtherAssignments() { |
| 150 |
List<SamplePairAssignmentForCell> result = new ArrayList<>(this.all); |
| 151 |
result.remove(this.best); |
| 152 |
return result; |
| 153 |
} |
| 154 |
|
| 155 |
/** |
| 156 |
* Returns the results for a pair of donors. If one of the donors given isn't in this data set, returns null. |
| 157 |
* @param donorOne |
| 158 |
* @param donorTwo |
| 159 |
* @return |
| 160 |
*/ |
| 161 |
public SamplePairAssignmentForCell getAssignmentForDonorPair(final String donorOne, final String donorTwo) { |
| 162 |
for (SamplePairAssignmentForCell o: this.all) |
| 163 |
if (o.getSampleOne().equals(donorOne) && o.getSampleTwo().equals(donorTwo)) |
| 164 |
return o; |
| 165 |
return null; |
| 166 |
} |
| 167 |
|
| 168 |
|
| 169 |
|
| 170 |
|
| 171 |
} |