net.maizegenetics.pangenome.smallseq.CreateSmallGenomesPlugin Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of phg Show documentation
Show all versions of phg Show documentation
PHG - Practical Haplotype Graph
/**
*
*/
package net.maizegenetics.pangenome.smallseq;
import java.awt.Frame;
import java.io.File;
import java.io.IOException;
import java.nio.file.Files;
import java.nio.file.Paths;
import java.util.stream.Collectors;
import javax.swing.ImageIcon;
import net.maizegenetics.pangenome.pipeline.MakeDefaultDirectoryPlugin;
import org.apache.commons.io.FileUtils;
import org.apache.log4j.Logger;
import net.maizegenetics.plugindef.AbstractPlugin;
import net.maizegenetics.plugindef.DataSet;
import net.maizegenetics.plugindef.GeneratePluginCode;
import net.maizegenetics.plugindef.PluginParameter;
import net.maizegenetics.util.LoggingUtils;
import static net.maizegenetics.pangenome.smallseq.SmallSeqPaths.*;
/**
* This class provides a standalone plugin to create the small sequence test genome files.
* Files are stored as per relative paths defined in SmallSeqPaths.java
*
* This plugin also creates reference index files as may be neede by BWA, GATK or other
* methods.
*
* Currently there are no required parameters - each parameter has a default.
*
* NOTE: This plugin creates the indices needed, so if the user is not running through
* RUnSmallSeqTestsDocker, the path to samtools, bwa and picard must be given.
* If running from a CBSU machine, these are:
*
* /programs/bin/bwa/bwa
* /programs/bin/picard-tools/picard
* /programs/bin/samtools/samtools
*
* To run this plugin on a cbsu machine using default parameters except for the indexes tools,
* do this (replace your own
*
* @author lcj34
*
*/
public class CreateSmallGenomesPlugin extends AbstractPlugin {
private static final Logger myLogger = Logger.getLogger(CreateSmallGenomesPlugin.class);
private PluginParameter geneLength = new PluginParameter.Builder("geneLength", 256, Integer.class).guiName("Length Of Genes").required(false)
.description("Number of base pairs to include for each gene region").build();
private PluginParameter interGeneLength = new PluginParameter.Builder("interGeneLength", 256, Integer.class).guiName("Length Of Inter Genes")
.description("Number of base pairs to include for each intergenic region").build();
private PluginParameter numberOfGenes = new PluginParameter.Builder("numberOfGenes", 3, Integer.class).guiName("Number Of Genes").required(false)
.description("Number of gene regions to create").build();
private PluginParameter readLength = new PluginParameter.Builder("readLength", 128, Integer.class).guiName("Length Of Reads").required(false)
.description("Desired length of reads for fastq files").build();
private PluginParameter hapDivergence = new PluginParameter.Builder("hapDivergence", 64, Integer.class).guiName("Haplotype Divergence").required(false)
.description("Percent each haplotype sequence should diverge from reference").build();
private PluginParameter interHapDivergence = new PluginParameter.Builder("interHapDivergence", 256, Integer.class).guiName("Inter-Haplotype Divergence").required(false)
.description("Percent each haplotype sequence should diverge from reference").build();
private PluginParameter refInterGeneDup = new PluginParameter.Builder("refInterGeneDup", 0.0, Double.class).guiName("Proportion of InterGene Duplicate").required(false)
.description("Proportion of the refererence inter-gene sequence to duplicate").build();
private PluginParameter refInterGeneDelete = new PluginParameter.Builder("refInterGeneDelete", 0.0, Double.class).guiName("Proportion of InterGene Deleted").required(false)
.description("Proportion of the reference inter-gene sequence to delete").build();
private PluginParameter refGeneInsert = new PluginParameter.Builder("refGeneInsert", 0.0, Double.class).guiName("Proportion of Gene With Insertions").required(false)
.description("Proportion of Gene where an insertion will be placed.").build();
private PluginParameter wgsDepth = new PluginParameter.Builder("wgsDepth", 10.0, Double.class).guiName("WGS Depth").required(false)
.description("Depth for WGS ").build();
private PluginParameter gbsDepth = new PluginParameter.Builder("gbsDepth", 0.5, Double.class).guiName("GBS Depth").required(false)
.description("Depth for GBS").build();
private PluginParameter baseDirectory = new PluginParameter.Builder("baseDir", null, String.class)
.description("The base directory for the small genomes data.").build();
@Override
public DataSet processData(DataSet input) {
Long totalTime = System.nanoTime();
LoggingUtils.setupDebugLogging();
if (baseDirectory() != null) SmallSeqPaths.resetBaseDirectory(baseDirectory());
try {
// Delete answerDir and baseDir then create new paths
// if SmallSeqPaths.resetBaseDirectory is run SmallSeqPaths.smallSeqBaseDir is set to baseDirectory()
// if we erased all the contents of baseDirectory() a user who did not realize that was going to happen
// might be moderately annoyed if it contained important files.
// As a result only subdirectories of smallSeqBaseDir created by this plugin are erased.
// If it does not exist smallSeqBaseDir is created by Files.createDirectories(Paths.get(baseDir))
if (Files.exists(Paths.get(baseDir))) FileUtils.deleteDirectory(new File(baseDir));
if (Files.exists(Paths.get(answerDir))) FileUtils.deleteDirectory(new File(answerDir));
Files.createDirectories(Paths.get(baseDir));
//Create DefaultDirectory
new MakeDefaultDirectoryPlugin()
.workingDir(baseDir)
.performFunction(null);
Files.createDirectories(Paths.get(alignDir));
Files.createDirectories(Paths.get(pangenomeDir));
myLogger.info("calling CreateTestGenomes.create to make the data sets");
// Begin data set creation.
CreateTestGenomes.create(geneLength(),interGeneLength(), refInterGeneDup(), refInterGeneDelete(),numberOfGenes(),
readLength(),hapDivergence(), interHapDivergence(), wgsDepth(), gbsDepth(), refGeneInsert());
} catch (IOException exc) {
System.err.println("Errors in creating genomes");
exc.printStackTrace();
}
return null;
}
public CreateSmallGenomesPlugin() {
super(null, false);
}
public CreateSmallGenomesPlugin(Frame parentFrame) {
super(parentFrame, false);
}
public CreateSmallGenomesPlugin(Frame parentFrame, boolean isInteractive) {
super(parentFrame, isInteractive);
}
@Override
public ImageIcon getIcon() {
return null;
}
@Override
public String getButtonName() {
return ("CreateTestGenomes");
}
@Override
public String getToolTipText() {
return ("Creates a set of small sequence genomes for testing with the PHG");
}
// The following getters and setters were auto-generated.
// Please use this method to re-generate.
//
public static void main(String[] args) {
GeneratePluginCode.generate(CreateSmallGenomesPlugin.class);
}
/**
* Number of base pairs to include for each gene region
*
* @return Length Of Genes
*/
public Integer geneLength() {
return geneLength.value();
}
/**
* Set Length Of Genes. Number of base pairs to include
* for each gene region
*
* @param value Length Of Genes
*
* @return this plugin
*/
public CreateSmallGenomesPlugin geneLength(Integer value) {
geneLength = new PluginParameter<>(geneLength, value);
return this;
}
/**
* Number of base pairs to include for each intergenic
* region
*
* @return Length Of Inter Genes
*/
public Integer interGeneLength() {
return interGeneLength.value();
}
/**
* Set Length Of Inter Genes. Number of base pairs to
* include for each intergenic region
*
* @param value Length Of Inter Genes
*
* @return this plugin
*/
public CreateSmallGenomesPlugin interGeneLength(Integer value) {
interGeneLength = new PluginParameter<>(interGeneLength, value);
return this;
}
/**
* Number of gene regions to create
*
* @return Number Of Genes
*/
public Integer numberOfGenes() {
return numberOfGenes.value();
}
/**
* Set Number Of Genes. Number of gene regions to create
*
* @param value Number Of Genes
*
* @return this plugin
*/
public CreateSmallGenomesPlugin numberOfGenes(Integer value) {
numberOfGenes = new PluginParameter<>(numberOfGenes, value);
return this;
}
/**
* Desired length of reads for fastq files
*
* @return Length Of Reads
*/
public Integer readLength() {
return readLength.value();
}
/**
* Set Length Of Reads. Desired length of reads for fastq
* files
*
* @param value Length Of Reads
*
* @return this plugin
*/
public CreateSmallGenomesPlugin readLength(Integer value) {
readLength = new PluginParameter<>(readLength, value);
return this;
}
/**
* Percent each haplotype sequence should diverge from
* reference
*
* @return Haplotype Divergence
*/
public Integer hapDivergence() {
return hapDivergence.value();
}
/**
* Set Haplotype Divergence. Percent each haplotype sequence
* should diverge from reference
*
* @param value Haplotype Divergence
*
* @return this plugin
*/
public CreateSmallGenomesPlugin hapDivergence(Integer value) {
hapDivergence = new PluginParameter<>(hapDivergence, value);
return this;
}
/**
* Percent each haplotype sequence should diverge from
* reference
*
* @return Inter-Haplotype Divergence
*/
public Integer interHapDivergence() {
return interHapDivergence.value();
}
/**
* Set Inter-Haplotype Divergence. Percent each haplotype
* sequence should diverge from reference
*
* @param value Inter-Haplotype Divergence
*
* @return this plugin
*/
public CreateSmallGenomesPlugin interHapDivergence(Integer value) {
interHapDivergence = new PluginParameter<>(interHapDivergence, value);
return this;
}
/**
* Proportion of the refererence inter-gene sequence to
* duplicate
*
* @return Proportion of InterGene Duplicate
*/
public Double refInterGeneDup() {
return refInterGeneDup.value();
}
/**
* Set Proportion of InterGene Duplicate. Proportion of
* the refererence inter-gene sequence to duplicate
*
* @param value Proportion of InterGene Duplicate
*
* @return this plugin
*/
public CreateSmallGenomesPlugin refInterGeneDup(Double value) {
refInterGeneDup = new PluginParameter<>(refInterGeneDup, value);
return this;
}
/**
* Proportion of the reference inter-gene sequence to
* delete
*
* @return Proportion of InterGene Deleted
*/
public Double refInterGeneDelete() {
return refInterGeneDelete.value();
}
/**
* Set Proportion of InterGene Deleted. Proportion of
* the reference inter-gene sequence to delete
*
* @param value Proportion of InterGene Deleted
*
* @return this plugin
*/
public CreateSmallGenomesPlugin refInterGeneDelete(Double value) {
refInterGeneDelete = new PluginParameter<>(refInterGeneDelete, value);
return this;
}
/**
* Proportion of gene that will get an insertion
*
* @return Proportion of InterGene Deleted
*/
public Double refGeneInsert() {
return refGeneInsert.value();
}
/**
* Set Proportion of gene that will get an insertion
*
* @param value length of insertions
*
* @return this plugin
*/
public CreateSmallGenomesPlugin refGeneInsert(Double value) {
refGeneInsert = new PluginParameter<>(refGeneInsert, value);
return this;
}
/**
* Depth for WGS
*
* @return WGS Depth
*/
public Double wgsDepth() {
return wgsDepth.value();
}
/**
* Set WGS Depth. Depth for WGS
*
* @param value WGS Depth
*
* @return this plugin
*/
public CreateSmallGenomesPlugin wgsDepth(Double value) {
wgsDepth = new PluginParameter<>(wgsDepth, value);
return this;
}
/**
* Depth for GBS
*
* @return GBS Depth
*/
public Double gbsDepth() {
return gbsDepth.value();
}
/**
* Set GBS Depth. Depth for GBS
*
* @param value GBS Depth
*
* @return this plugin
*/
public CreateSmallGenomesPlugin gbsDepth(Double value) {
gbsDepth = new PluginParameter<>(gbsDepth, value);
return this;
}
/**
* The base directory for the small genomes data.
*
* @return Base Dir
*/
public String baseDirectory() {
if (baseDirectory.value() == null) return null;
if (baseDirectory.value().endsWith("/")) return baseDirectory.value();
return baseDirectory.value() + "/";
}
/**
* Set Base Dir. The base directory for the small genomes
* data.
*
* @param value Base Dir
*
* @return this plugin
*/
public CreateSmallGenomesPlugin baseDirectory(String value) {
baseDirectory = new PluginParameter<>(baseDirectory, value);
return this;
}
}