formats: add FastQC support from opencga, #187

biodata-formats/src/main/java/org/opencb/biodata/formats/sequence/fastqc/AdapterContent.java

@@ -0,0 +1,91 @@
+package org.opencb.biodata.formats.sequence.fastqc;
+
+public class AdapterContent {
+// #Position	Illumina Universal Adapter	Illumina Small RNA 3' Adapter	Illumina Small RNA 5' Adapter	Nextera Transposase Sequence	SOLID Small RNA Adapter
+
+    private String position;
+    private double illumina;
+    private double illumina3;
+    private double illumina5;
+    private double nextera;
+    private double solid;
+
+    public AdapterContent() {
+    }
+
+    public AdapterContent(String position, double illumina, double illumina3, double illumina5, double nextera, double solid) {
+        this.position = position;
+        this.illumina = illumina;
+        this.illumina3 = illumina3;
+        this.illumina5 = illumina5;
+        this.nextera = nextera;
+        this.solid = solid;
+    }
+
+    @Override
+    public String toString() {
+        final StringBuilder sb = new StringBuilder("AdapterContent{");
+        sb.append("position='").append(position).append('\'');
+        sb.append(", illumina=").append(illumina);
+        sb.append(", illumina3=").append(illumina3);
+        sb.append(", illumina5=").append(illumina5);
+        sb.append(", nextera=").append(nextera);
+        sb.append(", solid=").append(solid);
+        sb.append('}');
+        return sb.toString();
+    }
+
+    public String getPosition() {
+        return position;
+    }
+
+    public AdapterContent setPosition(String position) {
+        this.position = position;
+        return this;
+    }
+
+    public double getIllumina() {
+        return illumina;
+    }
+
+    public AdapterContent setIllumina(double illumina) {
+        this.illumina = illumina;
+        return this;
+    }
+
+    public double getIllumina3() {
+        return illumina3;
+    }
+
+    public AdapterContent setIllumina3(double illumina3) {
+        this.illumina3 = illumina3;
+        return this;
+    }
+
+    public double getIllumina5() {
+        return illumina5;
+    }
+
+    public AdapterContent setIllumina5(double illumina5) {
+        this.illumina5 = illumina5;
+        return this;
+    }
+
+    public double getNextera() {
+        return nextera;
+    }
+
+    public AdapterContent setNextera(double nextera) {
+        this.nextera = nextera;
+        return this;
+    }
+
+    public double getSolid() {
+        return solid;
+    }
+
+    public AdapterContent setSolid(double solid) {
+        this.solid = solid;
+        return this;
+    }
+}

biodata-formats/src/main/java/org/opencb/biodata/formats/sequence/fastqc/FastQc.java

@@ -0,0 +1,194 @@
+package org.opencb.biodata.formats.sequence.fastqc;
+
+import java.util.*;
+
+public class FastQc {
+
+    private Summary summary;
+    private Map<String, String> basicStats;
+    private List<PerBaseSeqQuality> perBaseSeqQualities;
+    private Map<Integer, Double> perSeqQualityScores;
+    private List<PerTileSeqQuality> perTileSeqQuality;
+    private List<PerBaseSeqContent> perBaseSeqContent;
+    private double[] perSeqGcContent;
+    private Map<String, Double> perBaseNContent;
+    private Map<Integer, Double> seqLengthDistribution;
+    private List<SeqDuplicationLevel> seqDuplicationLevels;
+    private List<OverrepresentedSeq> overrepresentedSeqs;
+    private List<AdapterContent> adapterContent;
+    private List<KmerContent> kmerContent;
+
+    public FastQc() {
+        summary = new Summary();
+        basicStats = new LinkedHashMap<>();
+        perBaseSeqQualities = new LinkedList<>();
+        perSeqQualityScores = new LinkedHashMap<>();
+        perTileSeqQuality = new LinkedList<>();
+        perBaseSeqContent = new LinkedList<>();
+        perSeqGcContent = new double[101];
+        perBaseNContent = new LinkedHashMap<>();
+        seqLengthDistribution = new LinkedHashMap<>();
+        seqDuplicationLevels = new LinkedList<>();
+        overrepresentedSeqs = new LinkedList<>();
+        adapterContent = new LinkedList<>();
+        kmerContent = new LinkedList<>();
+    }
+
+    public FastQc(Summary summary, Map<String, String> basicStats, List<PerBaseSeqQuality> perBaseSeqQualities,
+                  Map<Integer, Double> perSeqQualityScores, List<PerTileSeqQuality> perTileSeqQuality,
+                  List<PerBaseSeqContent> perBaseSeqContent, double[] perSeqGcContent, Map<String, Double> perBaseNContent,
+                  Map<Integer, Double> seqLengthDistribution, List<SeqDuplicationLevel> seqDuplicationLevels,
+                  List<OverrepresentedSeq> overrepresentedSeqs, List<AdapterContent> adapterContent,
+                  List<KmerContent> kmerContent) {
+        this.summary = summary;
+        this.basicStats = basicStats;
+        this.perBaseSeqQualities = perBaseSeqQualities;
+        this.perSeqQualityScores = perSeqQualityScores;
+        this.perTileSeqQuality = perTileSeqQuality;
+        this.perBaseSeqContent = perBaseSeqContent;
+        this.perSeqGcContent = perSeqGcContent;
+        this.perBaseNContent = perBaseNContent;
+        this.seqLengthDistribution = seqLengthDistribution;
+        this.seqDuplicationLevels = seqDuplicationLevels;
+        this.overrepresentedSeqs = overrepresentedSeqs;
+        this.adapterContent = adapterContent;
+        this.kmerContent = kmerContent;
+    }
+
+    @Override
+    public String toString() {
+        final StringBuilder sb = new StringBuilder("FastQc{");
+        sb.append("summary=").append(summary);
+        sb.append(", basicStats=").append(basicStats);
+        sb.append(", perBaseSeqQualities=").append(perBaseSeqQualities);
+        sb.append(", perSeqQualityScores=").append(perSeqQualityScores);
+        sb.append(", perTileSeqQuality=").append(perTileSeqQuality);
+        sb.append(", perBaseSeqContent=").append(perBaseSeqContent);
+        sb.append(", perSeqGcContent=").append(Arrays.toString(perSeqGcContent));
+        sb.append(", perBaseNContent=").append(perBaseNContent);
+        sb.append(", seqLengthDistribution=").append(seqLengthDistribution);
+        sb.append(", seqDuplicationLevels=").append(seqDuplicationLevels);
+        sb.append(", overrepresentedSeqs=").append(overrepresentedSeqs);
+        sb.append(", adapterContent=").append(adapterContent);
+        sb.append(", kmerContent=").append(kmerContent);
+        sb.append('}');
+        return sb.toString();
+    }
+
+    public Summary getSummary() {
+        return summary;
+    }
+
+    public FastQc setSummary(Summary summary) {
+        this.summary = summary;
+        return this;
+    }
+
+    public Map<String, String> getBasicStats() {
+        return basicStats;
+    }
+
+    public FastQc setBasicStats(Map<String, String> basicStats) {
+        this.basicStats = basicStats;
+        return this;
+    }
+
+    public List<PerBaseSeqQuality> getPerBaseSeqQualities() {
+        return perBaseSeqQualities;
+    }
+
+    public FastQc setPerBaseSeqQualities(List<PerBaseSeqQuality> perBaseSeqQualities) {
+        this.perBaseSeqQualities = perBaseSeqQualities;
+        return this;
+    }
+
+    public Map<Integer, Double> getPerSeqQualityScores() {
+        return perSeqQualityScores;
+    }
+
+    public FastQc setPerSeqQualityScores(Map<Integer, Double> perSeqQualityScores) {
+        this.perSeqQualityScores = perSeqQualityScores;
+        return this;
+    }
+
+    public List<PerTileSeqQuality> getPerTileSeqQualities() {
+        return perTileSeqQuality;
+    }
+
+    public FastQc setPerTileSeqQuality(List<PerTileSeqQuality> perTileSeqQuality) {
+        this.perTileSeqQuality = perTileSeqQuality;
+        return this;
+    }
+
+    public List<PerBaseSeqContent> getPerBaseSeqContent() {
+        return perBaseSeqContent;
+    }
+
+    public FastQc setPerBaseSeqContent(List<PerBaseSeqContent> perBaseSeqContent) {
+        this.perBaseSeqContent = perBaseSeqContent;
+        return this;
+    }
+
+    public double[] getPerSeqGcContent() {
+        return perSeqGcContent;
+    }
+
+    public FastQc setPerSeqGcContent(double[] perSeqGcContent) {
+        this.perSeqGcContent = perSeqGcContent;
+        return this;
+    }
+
+    public Map<String, Double> getPerBaseNContent() {
+        return perBaseNContent;
+    }
+
+    public FastQc setPerBaseNContent(Map<String, Double> perBaseNContent) {
+        this.perBaseNContent = perBaseNContent;
+        return this;
+    }
+
+    public Map<Integer, Double> getSeqLengthDistribution() {
+        return seqLengthDistribution;
+    }
+
+    public FastQc setSeqLengthDistribution(Map<Integer, Double> seqLengthDistribution) {
+        this.seqLengthDistribution = seqLengthDistribution;
+        return this;
+    }
+
+    public List<SeqDuplicationLevel> getSeqDuplicationLevels() {
+        return seqDuplicationLevels;
+    }
+
+    public FastQc setSeqDuplicationLevels(List<SeqDuplicationLevel> seqDuplicationLevels) {
+        this.seqDuplicationLevels = seqDuplicationLevels;
+        return this;
+    }
+
+    public List<OverrepresentedSeq> getOverrepresentedSeqs() {
+        return overrepresentedSeqs;
+    }
+
+    public FastQc setOverrepresentedSeqs(List<OverrepresentedSeq> overrepresentedSeqs) {
+        this.overrepresentedSeqs = overrepresentedSeqs;
+        return this;
+    }
+
+    public List<AdapterContent> getAdapterContent() {
+        return adapterContent;
+    }
+
+    public FastQc setAdapterContent(List<AdapterContent> adapterContent) {
+        this.adapterContent = adapterContent;
+        return this;
+    }
+
+    public List<KmerContent> getKmerContent() {
+        return kmerContent;
+    }
+
+    public FastQc setKmerContent(List<KmerContent> kmerContent) {
+        this.kmerContent = kmerContent;
+        return this;
+    }
+}

biodata-formats/src/main/java/org/opencb/biodata/formats/sequence/fastqc/KmerContent.java

@@ -0,0 +1,79 @@
+package org.opencb.biodata.formats.sequence.fastqc;
+
+public class KmerContent {
+// #Sequence	Count	PValue	Obs/Exp Max	Max Obs/Exp Position
+
+    private String sequence;
+    private int count;
+    private double pValue;
+    private double obsExpMax;
+    private String maxObsExpPosition;
+
+    public KmerContent() {
+    }
+
+    public KmerContent(String sequence, int count, double pValue, double obsExpMax, String maxObsExpPosition) {
+        this.sequence = sequence;
+        this.count = count;
+        this.pValue = pValue;
+        this.obsExpMax = obsExpMax;
+        this.maxObsExpPosition = maxObsExpPosition;
+    }
+
+    @Override
+    public String toString() {
+        final StringBuilder sb = new StringBuilder("KmerContent{");
+        sb.append("sequence='").append(sequence).append('\'');
+        sb.append(", count=").append(count);
+        sb.append(", pValue=").append(pValue);
+        sb.append(", obsExpMax=").append(obsExpMax);
+        sb.append(", maxObsExpPosition='").append(maxObsExpPosition).append('\'');
+        sb.append('}');
+        return sb.toString();
+    }
+
+    public String getSequence() {
+        return sequence;
+    }
+
+    public KmerContent setSequence(String sequence) {
+        this.sequence = sequence;
+        return this;
+    }
+
+    public int getCount() {
+        return count;
+    }
+
+    public KmerContent setCount(int count) {
+        this.count = count;
+        return this;
+    }
+
+    public double getpValue() {
+        return pValue;
+    }
+
+    public KmerContent setpValue(double pValue) {
+        this.pValue = pValue;
+        return this;
+    }
+
+    public double getObsExpMax() {
+        return obsExpMax;
+    }
+
+    public KmerContent setObsExpMax(double obsExpMax) {
+        this.obsExpMax = obsExpMax;
+        return this;
+    }
+
+    public String getMaxObsExpPosition() {
+        return maxObsExpPosition;
+    }
+
+    public KmerContent setMaxObsExpPosition(String maxObsExpPosition) {
+        this.maxObsExpPosition = maxObsExpPosition;
+        return this;
+    }
+}