efischer/BiGpairSEQ
1
0
Fork 0
Code Issues Pull Requests Projects Releases 13 Wiki Activity

implement Zipf distribution

Browse Source
This commit is contained in:
eugenefischer
2025-04-09 14:32:02 -05:00
parent 161a52aa89
commit a43ee469ea
8 changed files with 169 additions and 54 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
.idea/.name generated Normal file
View File

@@ -0,0 +1 @@
BiGpairSEQ

13
.idea/libraries/commons_rng_1.xml generated Normal file
View File

@@ -0,0 +1,13 @@
<component name="libraryTable">
<library name="commons-rng-1">
<CLASSES>
<root url="file://$USER_HOME$/Downloads/commons-rng-1.6" />
</CLASSES>
<JAVADOC />
<SOURCES>
<root url="file://$USER_HOME$/Downloads/commons-rng-1.6" />
</SOURCES>
<jarDirectory url="file://$USER_HOME$/Downloads/commons-rng-1.6" recursive="false" />
<jarDirectory url="file://$USER_HOME$/Downloads/commons-rng-1.6" recursive="false" type="SOURCES" />
</library>
</component>

5
src/main/java/BiGpairSEQ.java
View File

@@ -15,6 +15,7 @@ public class BiGpairSEQ {
private static boolean cacheGraph = false; private static boolean cacheGraph = false;
private static AlgorithmType matchingAlgoritmType = AlgorithmType.HUNGARIAN; private static AlgorithmType matchingAlgoritmType = AlgorithmType.HUNGARIAN;
private static HeapType priorityQueueHeapType = HeapType.PAIRING; private static HeapType priorityQueueHeapType = HeapType.PAIRING;
private static DistributionType distributionType = DistributionType.ZIPF;
private static boolean outputBinary = true; private static boolean outputBinary = true;
private static boolean outputGraphML = false; private static boolean outputGraphML = false;
private static boolean calculatePValue = false; private static boolean calculatePValue = false;
@@ -60,6 +61,10 @@ public class BiGpairSEQ {
return cellFilename; return cellFilename;
} }
public static DistributionType getDistributionType() {return distributionType;}
public static void setDistributionType(DistributionType type) {distributionType = type;}
public static Plate getPlateInMemory() { public static Plate getPlateInMemory() {
return plateInMemory; return plateInMemory;
} }

19
src/main/java/CommandLineInterface.java
View File

@@ -123,16 +123,20 @@ public class CommandLineInterface {
Plate plate; Plate plate;
if (line.hasOption("poisson")) { if (line.hasOption("poisson")) {
Double stdDev = Math.sqrt(numWells); Double stdDev = Math.sqrt(numWells);
plate = new Plate(cells, cellFilename, numWells, populations, dropoutRate, stdDev, false); plate = new Plate(cells, cellFilename, numWells, populations, dropoutRate, stdDev);
} }
else if (line.hasOption("gaussian")) { else if (line.hasOption("gaussian")) {
Double stdDev = Double.parseDouble(line.getOptionValue("stddev")); Double stdDev = Double.parseDouble(line.getOptionValue("stddev"));
plate = new Plate(cells, cellFilename, numWells, populations, dropoutRate, stdDev, false); plate = new Plate(cells, cellFilename, numWells, populations, dropoutRate, stdDev);
}
else if (line.hasOption("zipf")) {
Double zipfExponent = Double.parseDouble(line.getOptionValue("exp"));
plate = new Plate(cells, cellFilename, numWells, populations, dropoutRate, zipfExponent);
} }
else { else {
assert line.hasOption("exponential"); assert line.hasOption("exponential");
Double lambda = Double.parseDouble(line.getOptionValue("lambda")); Double lambda = Double.parseDouble(line.getOptionValue("lambda"));
plate = new Plate(cells, cellFilename, numWells, populations, dropoutRate, lambda, true); plate = new Plate(cells, cellFilename, numWells, populations, dropoutRate, lambda);
} }
PlateFileWriter writer = new PlateFileWriter(outputFilename, plate); PlateFileWriter writer = new PlateFileWriter(outputFilename, plate);
writer.writePlateFile(); writer.writePlateFile();
@@ -340,9 +344,13 @@ public class CommandLineInterface {
Option exponential = Option.builder("exponential") Option exponential = Option.builder("exponential")
.desc("Use an exponential distribution for cell sample") .desc("Use an exponential distribution for cell sample")
.build(); .build();
Option zipf = Option.builder("zipf")
.desc("Use a Zipf distribution for cell sample")
.build();
distributions.addOption(poisson); distributions.addOption(poisson);
distributions.addOption(gaussian); distributions.addOption(gaussian);
distributions.addOption(exponential); distributions.addOption(exponential);
distributions.addOption(zipf);
//options group for statistical distribution parameters //options group for statistical distribution parameters
OptionGroup statParams = new OptionGroup();// add this to plate options OptionGroup statParams = new OptionGroup();// add this to plate options
Option stdDev = Option.builder("stddev") Option stdDev = Option.builder("stddev")
@@ -355,6 +363,11 @@ public class CommandLineInterface {
.hasArg() .hasArg()
.argName("value") .argName("value")
.build(); .build();
Option zipfExponent = Option.builder("exp")
.desc("If using -zipf flag, exponent value for distribution")
.hasArg()
.argName("value")
.build();
statParams.addOption(stdDev); statParams.addOption(stdDev);
statParams.addOption(lambda); statParams.addOption(lambda);
//Option group for random plate or set populations //Option group for random plate or set populations

6
src/main/java/DistributionType.java Normal file
View File

@@ -0,0 +1,6 @@
public enum DistributionType {
POISSON,
GAUSSIAN,
EXPONENTIAL,
ZIPF
}

63
src/main/java/InteractiveInterface.java
View File

@@ -89,14 +89,12 @@ public class InteractiveInterface {
private static void makePlate() { private static void makePlate() {
String cellFile = null; String cellFile = null;
String filename = null; String filename = null;
Double stdDev = 0.0; Double parameter = 0.0;
Integer numWells = 0; Integer numWells = 0;
Integer numSections; Integer numSections;
Integer[] populations = {1}; Integer[] populations = {1};
Double dropOutRate = 0.0; Double dropOutRate = 0.0;
boolean poisson = false; ;
boolean exponential = false;
double lambda = 1.5;
try { try {
System.out.println("\nSimulated sample plates consist of:"); System.out.println("\nSimulated sample plates consist of:");
System.out.println("* a number of wells"); System.out.println("* a number of wells");
@@ -114,33 +112,46 @@ public class InteractiveInterface {
System.out.println("1) Poisson"); System.out.println("1) Poisson");
System.out.println("2) Gaussian"); System.out.println("2) Gaussian");
System.out.println("3) Exponential"); System.out.println("3) Exponential");
// System.out.println("(Note: approximate distribution in original paper is exponential, lambda = 0.6)"); System.out.println("4) Zipf");
// System.out.println("(lambda value approximated from slope of log-log graph in figure 4c)");
System.out.println("(Note: wider distributions are more memory intensive to match)"); System.out.println("(Note: wider distributions are more memory intensive to match)");
System.out.print("Enter selection value: "); System.out.print("Enter selection value: ");
input = sc.nextInt(); input = sc.nextInt();
switch (input) { switch (input) {
case 1 -> poisson = true; case 1 -> {
BiGpairSEQ.setDistributionType(DistributionType.POISSON);
}
case 2 -> { case 2 -> {
BiGpairSEQ.setDistributionType(DistributionType.GAUSSIAN);
System.out.println("How many distinct T-cells within one standard deviation of peak frequency?"); System.out.println("How many distinct T-cells within one standard deviation of peak frequency?");
System.out.println("(Note: wider distributions are more memory intensive to match)"); System.out.println("(Note: wider distributions are more memory intensive to match)");
stdDev = sc.nextDouble(); parameter = sc.nextDouble();
if (stdDev <= 0.0) { if (parameter <= 0.0) {
throw new InputMismatchException("Value must be positive."); throw new InputMismatchException("Value must be positive.");
} }
} }
case 3 -> { case 3 -> {
exponential = true; BiGpairSEQ.setDistributionType(DistributionType.EXPONENTIAL);
System.out.print("Please enter lambda value for exponential distribution: "); System.out.print("Please enter lambda value for exponential distribution: ");
lambda = sc.nextDouble(); parameter = sc.nextDouble();
if (lambda <= 0.0) { if (parameter <= 0.0) {
lambda = 0.6; parameter = 1.4;
System.out.println("Value must be positive. Defaulting to 0.6."); System.out.println("Value must be positive. Defaulting to 1.4.");
}
}
case 4 -> {
BiGpairSEQ.setDistributionType(DistributionType.ZIPF);
System.out.print("Please enter exponent value for Zipf distribution: ");
parameter = sc.nextDouble();
if (parameter <= 0.0) {
parameter = 1.4;
System.out.println("Value must be positive. Defaulting to 1.4.");
} }
} }
default -> { default -> {
System.out.println("Invalid input. Defaulting to exponential."); System.out.println("Invalid input. Defaulting to exponential.");
exponential = true; parameter = 1.4;
BiGpairSEQ.setDistributionType(DistributionType.EXPONENTIAL);
} }
} }
System.out.print("\nNumber of wells on plate: "); System.out.print("\nNumber of wells on plate: ");
@@ -226,16 +237,17 @@ public class InteractiveInterface {
assert filename != null; assert filename != null;
Plate samplePlate; Plate samplePlate;
PlateFileWriter writer; PlateFileWriter writer;
if(exponential){ DistributionType type = BiGpairSEQ.getDistributionType();
samplePlate = new Plate(cells, cellFile, numWells, populations, dropOutRate, lambda, true); switch(type) {
writer = new PlateFileWriter(filename, samplePlate); case POISSON -> {
} parameter = Math.sqrt(cells.getCellCount()); //gaussian with square root of elements approximates poisson
else { samplePlate = new Plate(cells, cellFile, numWells, populations, dropOutRate, parameter);
if (poisson) { writer = new PlateFileWriter(filename, samplePlate);
stdDev = Math.sqrt(cells.getCellCount()); //gaussian with square root of elements approximates poisson }
default -> {
samplePlate = new Plate(cells, cellFile, numWells, populations, dropOutRate, parameter);
writer = new PlateFileWriter(filename, samplePlate);
} }
samplePlate = new Plate(cells, cellFile, numWells, populations, dropOutRate, stdDev, false);
writer = new PlateFileWriter(filename, samplePlate);
} }
System.out.println("Writing Sample Plate to file"); System.out.println("Writing Sample Plate to file");
writer.writePlateFile(); writer.writePlateFile();
@@ -605,12 +617,13 @@ public class InteractiveInterface {
case 3 -> { case 3 -> {
BiGpairSEQ.setAuctionAlgorithm(); BiGpairSEQ.setAuctionAlgorithm();
System.out.println("MWM algorithm set to auction"); System.out.println("MWM algorithm set to auction");
backToOptions = true;
} }
case 4 -> { case 4 -> {
System.out.println("Scaling integer weight MWM algorithm not yet fully implemented. Sorry."); System.out.println("Scaling integer weight MWM algorithm not yet fully implemented. Sorry.");
// BiGpairSEQ.setIntegerWeightScalingAlgorithm(); // BiGpairSEQ.setIntegerWeightScalingAlgorithm();
// System.out.println("MWM algorithm set to integer weight scaling algorithm of Duan and Su"); // System.out.println("MWM algorithm set to integer weight scaling algorithm of Duan and Su");
backToOptions = true; // backToOptions = true;
} }
case 0 -> backToOptions = true; case 0 -> backToOptions = true;
default -> System.out.println("Invalid input"); default -> System.out.println("Invalid input");

74
src/main/java/Plate.java
View File

@@ -13,6 +13,11 @@ TODO: Implement discrete frequency distributions using Vose's Alias Method
*/ */
import org.apache.commons.rng.UniformRandomProvider;
import org.apache.commons.rng.core.BaseProvider;
import org.apache.commons.rng.sampling.distribution.RejectionInversionZipfSampler;
import org.apache.commons.rng.simple.JDKRandomWrapper;
import java.util.*; import java.util.*;
public class Plate { public class Plate {
@@ -26,25 +31,22 @@ public class Plate {
private Integer[] populations; private Integer[] populations;
private double stdDev; private double stdDev;
private double lambda; private double lambda;
boolean exponential = false; private double zipfExponent;
private DistributionType distributionType;
public Plate(CellSample cells, String cellFilename, int numWells, Integer[] populations, public Plate(CellSample cells, String cellFilename, int numWells, Integer[] populations,
double dropoutRate, double stdDev_or_lambda, boolean exponential){ double dropoutRate, double parameter){
this.cells = cells; this.cells = cells;
this.sourceFile = cellFilename; this.sourceFile = cellFilename;
this.size = numWells; this.size = numWells;
this.wells = new ArrayList<>(); this.wells = new ArrayList<>();
this.error = dropoutRate; this.error = dropoutRate;
this.populations = populations; this.populations = populations;
this.exponential = exponential; this.stdDev = parameter;
if (this.exponential) { this.lambda = parameter;
this.lambda = stdDev_or_lambda; this.zipfExponent = parameter;
fillWellsExponential(cells.getCells(), this.lambda); this.distributionType = BiGpairSEQ.getDistributionType();
} fillWells(cells.getCells());
else {
this.stdDev = stdDev_or_lambda;
fillWells(cells.getCells(), this.stdDev);
}
} }
@@ -85,9 +87,33 @@ public class Plate {
} }
} }
private void fillWellsZipf(List<String[]> cells, double exponent) {
int numSections = populations.length;
int section = 0;
int n;
RejectionInversionZipfSampler zipfSampler = new RejectionInversionZipfSampler(new JDKRandomWrapper(rand), cells.size(), exponent);
while (section < numSections){
for (int i = 0; i < (size / numSections); i++) {
List<String[]> well = new ArrayList<>();
for (int j = 0; j < populations[section]; j++) {
do {
n = zipfSampler.sample();
} while (n >= cells.size() || n < 0);
String[] cellToAdd = cells.get(n).clone();
for(int k = 0; k < cellToAdd.length; k++){
if(Math.abs(rand.nextDouble()) < error){//error applied to each sequence
cellToAdd[k] = "-1";
}
}
well.add(cellToAdd);
}
wells.add(well);
}
section++;
}
}
private void fillWellsExponential(List<String[]> cells, double lambda){ private void fillWellsExponential(List<String[]> cells, double lambda){
this.lambda = lambda;
exponential = true;
int numSections = populations.length; int numSections = populations.length;
int section = 0; int section = 0;
double m; double m;
@@ -143,6 +169,24 @@ public class Plate {
} }
} }
private void fillWells(List<String[]> cells){
DistributionType type = BiGpairSEQ.getDistributionType();
switch (type) {
case POISSON, GAUSSIAN -> {
fillWells(cells, getStdDev());
break;
}
case EXPONENTIAL -> {
fillWellsExponential(cells, getLambda());
break;
}
case ZIPF -> {
fillWellsZipf(cells, getZipfExponent());
break;
}
}
}
public Integer[] getPopulations(){ public Integer[] getPopulations(){
return populations; return populations;
} }
@@ -155,10 +199,12 @@ public class Plate {
return stdDev; return stdDev;
} }
public boolean isExponential(){return exponential;} public DistributionType getDistributionType() { return distributionType;}
public double getLambda(){return lambda;} public double getLambda(){return lambda;}
public double getZipfExponent(){return zipfExponent;}
public double getError() { public double getError() {
return error; return error;
} }

42
src/main/java/PlateFileWriter.java
View File

@@ -13,11 +13,13 @@ public class PlateFileWriter {
private List<List<String[]>> wells; private List<List<String[]>> wells;
private double stdDev; private double stdDev;
private double lambda; private double lambda;
private double zipfExponent;
private DistributionType distributionType;
private Double error; private Double error;
private String filename; private String filename;
private String sourceFileName; private String sourceFileName;
private Integer[] populations; private Integer[] populations;
private boolean isExponential = false;
public PlateFileWriter(String filename, Plate plate) { public PlateFileWriter(String filename, Plate plate) {
if(!filename.matches(".*\\.csv")){ if(!filename.matches(".*\\.csv")){
@@ -26,12 +28,17 @@ public class PlateFileWriter {
this.filename = filename; this.filename = filename;
this.sourceFileName = plate.getSourceFileName(); this.sourceFileName = plate.getSourceFileName();
this.size = plate.getSize(); this.size = plate.getSize();
this.isExponential = plate.isExponential(); this.distributionType = plate.getDistributionType();
if(isExponential) { switch(distributionType) {
this.lambda = plate.getLambda(); case POISSON, GAUSSIAN -> {
} this.stdDev = plate.getStdDev();
else{ }
this.stdDev = plate.getStdDev(); case EXPONENTIAL -> {
this.lambda = plate.getLambda();
}
case ZIPF -> {
this.zipfExponent = plate.getZipfExponent();
}
} }
this.error = plate.getError(); this.error = plate.getError();
this.wells = plate.getWells(); this.wells = plate.getWells();
@@ -95,11 +102,22 @@ public class PlateFileWriter {
printer.printComment("Plate size: " + size); printer.printComment("Plate size: " + size);
printer.printComment("Well populations: " + wellPopulationsString); printer.printComment("Well populations: " + wellPopulationsString);
printer.printComment("Error rate: " + error); printer.printComment("Error rate: " + error);
if(isExponential){ switch (distributionType) {
printer.printComment("Lambda: " + lambda); case POISSON -> {
} printer.printComment("Cell frequency distribution: POISSON");
else { }
printer.printComment("Std. dev.: " + stdDev); case GAUSSIAN -> {
printer.printComment("Cell frequency distribution: GAUSSIAN");
printer.printComment("--Standard deviation: " + stdDev);
}
case EXPONENTIAL -> {
printer.printComment("Cell frequency distribution: EXPONENTIAL");
printer.printComment("--Lambda: " + lambda);
}
case ZIPF -> {
printer.printComment("Cell frequency distribution: ZIPF");
printer.printComment("--Exponent: " + zipfExponent);
}
} }
printer.printRecords(wellsAsStrings); printer.printRecords(wellsAsStrings);
} catch(IOException ex){ } catch(IOException ex){