Merge pull request #22 from apeltzer/master

V0.4.3

Author: apeltzer

.travis.yml

@@ -4,12 +4,12 @@ jdk:
 - oraclejdk8
 script: gradle build
 env:
-- version=0.4.2
+- version=0.4.3
 deploy:
   provider: releases
   api_key:
     secure: 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
-  file: build/libs/DamageProfiler-0.4.2.jar
+  file: build/libs/DamageProfiler-0.4.3.jar
   skip_cleanup: true
   on:
     all_branches: true

build.gradle

@@ -1,5 +1,5 @@
 group 'com.uni-tuebingen.de.it.eager.damageprofiler'
-version '0.4.2'
+version '0.4.3'
 
 
 
@@ -50,7 +50,7 @@ dependencies {
     compile group: 'com.intellij', name: 'forms_rt', version: '6.0.+'
     compile group: 'log4j', name: 'log4j', version: '1.+'
     compile group: 'com.github.broadinstitute', name: 'picard', version: '2.+'
-    compile 'org.yaml:snakeyaml:1.2+'
+    compile 'com.google.code.gson:gson:2.8.5'
 }
 
 

src/main/java/IO/OutputGenerator.java

@@ -2,17 +2,18 @@
 
 import calculations.DamageProfiler;
 import calculations.Frequencies;
+import com.google.gson.Gson;
 import com.itextpdf.awt.PdfGraphics2D;
 import com.itextpdf.text.*;
 import com.itextpdf.text.Font;
 import com.itextpdf.text.pdf.PdfContentByte;
 import com.itextpdf.text.pdf.PdfTemplate;
 import com.itextpdf.text.pdf.PdfWriter;
+import org.apache.commons.math3.stat.descriptive.DescriptiveStatistics;
 import org.apache.log4j.Logger;
 import org.jfree.chart.JFreeChart;
 import org.jfree.data.statistics.HistogramDataset;
 import org.jfree.data.xy.XYDataset;
-import org.yaml.snakeyaml.Yaml;
 
 
 import java.awt.*;
@@ -43,6 +44,7 @@ public class OutputGenerator {
     private int threshold;
     private int length;
     private String input;
+    private HashMap<String,Object> json_map = new HashMap<>();
 
 
 
@@ -65,6 +67,37 @@ public OutputGenerator(String outputFolder, DamageProfiler damageProfiler, Strin
         }
     }
 
+    /**
+     * writes all generated output statistics to YAML output file
+     * YAML has several key,value pairs (HashMaps) that all contain information created in DamageProfiler
+     * sample_name = the name of the sample
+     *
+     *
+     * @throws IOException
+     */
+    public void writeJSON(String version) throws IOException {
+        //Add Sample Name to yaml
+        String sampleName = input.split("/")[input.split("/").length-1];
+
+
+        //Add Metadata to JSON output
+        HashMap<String, Object> meta_map = new HashMap<>();
+        meta_map.put("sample_name", sampleName);
+        meta_map.put("tool_name", "DamageProfiler");
+        meta_map.put("version", version);
+
+        json_map.put("metadata", meta_map);
+
+        Gson gson = new Gson();
+
+        String json = gson.toJson(json_map);
+
+        FileWriter fw = new FileWriter(this.outpath + "/dmgprof.json");
+        fw.write(json);
+        fw.flush();
+        fw.close();
+
+    }
 
     /**
      * create tab-separated txt file with all read length, sorted according strand direction
@@ -73,12 +106,7 @@ public OutputGenerator(String outputFolder, DamageProfiler damageProfiler, Strin
      */
     public void writeLengthDistribution() throws IOException{
 
-
         BufferedWriter lgdist = new BufferedWriter(new FileWriter(this.outpath + "/lgdistribution.txt"));
-        Yaml lgdistyaml_fw = new Yaml();
-        Yaml lgdistyaml_rv = new Yaml();
-        FileWriter writer_fw = new FileWriter(this.outpath + "/lgdistribution_fw.yaml");
-        FileWriter writer_rv = new FileWriter(this.outpath + "/lgdistribution_rv.yaml");
         HashMap<Integer, Integer> map_forward = damageProfiler.getLength_distribution_map_forward();
         HashMap<Integer, Integer> map_reverse = damageProfiler.getLength_distribution_map_reverse();
 
@@ -94,8 +122,6 @@ public void writeLengthDistribution() throws IOException{
         key_list.addAll(map_forward.keySet());
         Collections.sort(key_list);
 
-        HashMap<String,HashMap> dump_fw_hm = new HashMap<>();
-        String sampleid= input.split("/")[input.split("/").length-1];
         HashMap<Integer,Integer> yaml_dump_fw = new HashMap<>();
 
         if(key_list.size()>0){
@@ -107,15 +133,13 @@ public void writeLengthDistribution() throws IOException{
                 yaml_dump_fw.put(key, map_forward.get(key));
             }
         }
-        dump_fw_hm.put(sampleid,yaml_dump_fw);
-        lgdistyaml_fw.dump(dump_fw_hm,writer_fw);
+        json_map.put("lendist_fw",yaml_dump_fw);
 
         key_list.clear();
         key_list.addAll(map_reverse.keySet());
         Collections.sort(key_list);
 
         HashMap<Integer,Integer> yaml_dump_rv = new HashMap<>();
-        HashMap<String,HashMap> dump_rv_hm = new HashMap<>();
 
 
         if(key_list.size()>0){
@@ -133,12 +157,7 @@ public void writeLengthDistribution() throws IOException{
 
         }
 
-        dump_rv_hm.put(sampleid,yaml_dump_rv);
-
-        lgdistyaml_rv.dump(dump_rv_hm,writer_rv);
-
-
-
+        json_map.put("lendist_rv", yaml_dump_rv);
         lgdist.close();
     }
 
@@ -453,22 +472,10 @@ public void writeDamageFiles(double[] gToA_reverse, double[] cToT) throws IOExce
         BufferedWriter writer3Prime = new BufferedWriter(new FileWriter(this.outpath + "/3pGtoA_freq.txt"));
         BufferedWriter writer5Prime = new BufferedWriter(new FileWriter(this.outpath + "/5pCtoT_freq.txt"));
 
-        FileWriter writ3P = new FileWriter(this.outpath + "/3pGtoA_freq.yaml");
-        FileWriter writ5p = new FileWriter(this.outpath + "/5pCtoT_freq.yaml");
+        //Add stuff to json output file
+        json_map.put("dmg_5p",getSubArray(cToT, this.threshold));
+        json_map.put("dmg_3p",getSubArray(gToA_reverse,this.threshold));
 
-        String sampleid= input.split("/")[input.split("/").length-1];
-
-        Yaml yml3p = new Yaml();
-        Yaml yml5p = new Yaml();
-
-        Map<String, double[]> yml3p_map = new HashMap<String,double[]>();
-        yml3p_map.put(sampleid,getSubArray(cToT, this.threshold));
-        yml3p.dump(yml3p_map,writ3P);
-
-
-        Map<String, double[]> yml5p_map = new HashMap<String,double[]>();
-        yml5p_map.put(sampleid,getSubArray(gToA_reverse,this.threshold));
-        yml5p.dump(yml5p_map,writ5p);
 
         writer3Prime.write("# table produced by calculations.DamageProfiler\n");
         writer3Prime.write("# using mapped file " + input + "\n");
@@ -502,6 +509,41 @@ private void writeDamagePattern(String title, double[] values, BufferedWriter wr
 
     }
 
+    public void computeSummaryMetrics(){
+        HashMap<Integer,Integer> forwardMap = damageProfiler.getLength_distribution_map_forward(); // key = length, value = occurences
+        HashMap<Integer, Integer>reverseMap = damageProfiler.getLength_distribution_map_reverse();
+
+        //Create ArrayList<Integer> of read lengths
+        DescriptiveStatistics descriptiveStatistics = new DescriptiveStatistics();
+
+        for (int key : forwardMap.keySet()){
+            int occurences = forwardMap.get(key);
+            for (int i = 0; i <= occurences; i++) {
+                descriptiveStatistics.addValue(key);
+            }
+        }
+
+        for (int key : reverseMap.keySet()){
+            int occurences = reverseMap.get(key);
+            for (int i = 0; i <= occurences; i++) {
+                descriptiveStatistics.addValue(key);
+            }
+        }
+
+        double mean = descriptiveStatistics.getMean();
+        double std = descriptiveStatistics.getStandardDeviation();
+        double median = descriptiveStatistics.getPercentile(50);
+
+        HashMap<String, Double> summ_stats = new HashMap<>();
+
+        summ_stats.put("mean_readlength", mean);
+        summ_stats.put("std", std);
+        summ_stats.put("median", std);
+
+        json_map.put("summary_stats", summ_stats);
+
+    }
+
 
 
     public void writeMisincorporations(Frequencies frequencies, int threshold) throws IOException {

src/main/java/RunDamageProfiler.java

@@ -9,7 +9,7 @@
  */
 public class RunDamageProfiler {
 
-    private static final String VERSION = "0.3.14";
+    private static final String VERSION = "0.4.3";
 
 
     @SuppressWarnings("static-access")

src/main/java/calculations/StartCalculations.java

@@ -35,6 +35,7 @@ public class StartCalculations {
     private String input;
     private SpecieHandler specieHandler;
 
+
     public StartCalculations(String version){
         VERSION = version;
     }
@@ -280,7 +281,6 @@ parse species reference (-s) and run DP
 
     }
 
-
     private void generateOutput(DamageProfiler damageProfiler, String output_folder, String inputfileNameWithOutExtension, String spe)
             throws IOException, DocumentException {
 
@@ -296,6 +296,7 @@ private void generateOutput(DamageProfiler damageProfiler, String output_folder,
                     input,
                     LOG
             );
+
             outputGenerator.writeLengthDistribution();
             outputGenerator.writeDamageFiles(
                     damageProfiler.getFrequencies().getCount_G_A_3_norm(),
@@ -308,6 +309,8 @@ private void generateOutput(DamageProfiler damageProfiler, String output_folder,
                     damageProfiler.getFrequencies(),
                     threshold
             );
+            outputGenerator.computeSummaryMetrics();
+            outputGenerator.writeJSON(VERSION);
 
 
             // create DamagePlots of 3' and 5' ends