vm version

SETUPINSTR.md

@@ -10,13 +10,17 @@ Enhort is build to run on two different servers, a computation back-end server a
 # Database 
 
 ### Use pre-build database
-Download the SQlite database and the corresponding .bed tracks from .... Save the database and the data files in a known directory.
+Download the minimal SQlite database and the corresponding .bed-tracks from 
+https://homepages.thm.de/~mmnz21/minimal.db 
+https://homepages.thm.de/~mmnz21/enhort_bed_files_hg19.tar.gz %add hg19 folder
 
-### Build own database
+Save the database and unpack .bed-files in a known directory.
+
+### Build your own database
 TODO 
 
 # Server setup
-Download the enhort.jar from ....
+Download the enhort.jar from https://homepages.thm.de/~mmnz21/enhort.jar
 
 Create a empty directory /logs in user home directory for logfiles
 
@@ -33,7 +37,8 @@ The server is run with the following command:
 
 
 # Frontend setup
-Download the frontend archive containing the .jar and the contig sizes for hg19 and hg38.
+Download the frontend archive containing the .jar and the contig sizes for hg19 and hg38 from 
+https://homepages.thm.de/~mmnz21/frontend.tar.gz
 Create a empty directory /logs in user home directory for logfiles
 
 The frontend is run with the following command:
@@ -53,6 +58,17 @@ The frontend is run with the following command:
 TODO 
 
 
+### Package Frontend.tar.gz
+ tar cvf frontend.tar frontend.jar ../../con/
+
+####  Content: 
+-  5 frontend.jar
+-  6 con/
+-  7 con/contigs_GRCh38
+-  8 con/contigs_hg19
+-  9 con/contigs_hg18                                                                                                
+
+
 ### Known issues
 
 - The sample button does not work on custom systems because the .bed-file used for the sample run is not available

backend/src/main/java/de/thm/genomeData/tracks/FileLoader.java

@@ -98,6 +98,7 @@ class FileLoader implements Callable<Optional<Track>> {
         List<String> names = new ArrayList<>(linecount);
         List<Double> scores = new ArrayList<>(linecount);
         List<Character> strands = new ArrayList<>(linecount);
+        int counter = 0;
 
         try (Stream<String> lines = Files.lines(file.toPath(), StandardCharsets.UTF_8)) {
             Iterator<String> it = lines.iterator();
@@ -124,6 +125,7 @@ class FileLoader implements Callable<Optional<Track>> {
 
                 if (line_matcher.matches()) {
                     String[] parts = line.split("\t");
+                    counter++;
 
                     long start;
                     long end;
@@ -186,10 +188,16 @@ class FileLoader implements Callable<Optional<Track>> {
                         else strands.add('o');
                     }
                 }
+
             }
 
             lines.close();
 
+
+            if (linecount != counter)
+                logger.error("File " + file.getName() + " has " + counter + " lines, but db says it has " + linecount + " lines.");
+
+
            // Check read files //
 
             if (logger.isDebugEnabled()) {

backend/src/main/java/de/thm/genomeData/tracks/TrackFactory.java

@@ -23,6 +23,9 @@ import de.thm.misc.Genome;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
+import java.io.IOException;
+import java.nio.file.Files;
+import java.nio.file.Paths;
 import java.util.*;
 import java.util.concurrent.*;
 import java.util.stream.Collectors;
@@ -108,12 +111,14 @@ public final class TrackFactory {
 
         if (System.getenv("HOME").contains("menzel")) {
 
+            allTracks = connector.getAllTracks("WHERE name like '%ontigs'");
+            /*
             allTracks = connector.getAllTracks(" WHERE directory like '%genetic%' and genome_assembly = 'hg19' ORDER BY lines ASC ");
             allTracks.addAll(connector.getAllTracks("WHERE genome_assembly = 'hg19' and directory like '%encode%HeLa%'"));
             allTracks.addAll(connector.getAllTracks("WHERE genome_assembly = 'hg19' and directory like '%custom%'"));
             //allTracks = connector.getAllTracks("WHERE bed_filename = 'SRX062365.05.bed'");
-            allTracks.addAll(connector.getAllTracks("WHERE name like '%ontigs'"));
             //allTracks = connector.getAllTracks(" WHERE genome_assembly = 'hg19' ORDER BY lines ASC LIMIT 3000");
+            */
         } else {
 
             allTracks = connector.getAllTracks("WHERE (cell_line NOT like '%GM%' or cell_line like '%GM12878')");
@@ -132,6 +137,27 @@ public final class TrackFactory {
         logger.info("Trying to load " + allTracks.size() + " tracks");
         this.tracks.addAll(loadByEntries(allTracks));
 
+        List<TrackEntry> customTracks = new ArrayList<>();
+
+        // Add custom tracks
+        if (System.getenv("HOME").contains("menzel")) {
+            try {
+
+                Files.walk(Paths.get("/home/menzel/Desktop/THM/promotion/enhort/dat/stefan/custom"))
+               .filter(Files::isRegularFile)
+               .filter(f -> f.getFileName().toString().endsWith(".bed"))
+               .forEach(f -> customTracks.add(new TrackEntry(f.getFileName().toString(),"Custom track: " + f.getFileName(),
+                        "/home/menzel/Desktop/THM/promotion/enhort/dat/stefan/custom/" + f.getFileName().toString(), "inout", Genome.Assembly.GRCh38.toString(), "Unknown", 14003120, "Genetic", 1412123+f.getFileName().hashCode(), "nosource", "nourl")));
+            } catch (IOException e) {
+                e.printStackTrace();
+            }
+        }
+
+        logger.info("Trying to load " + customTracks.size() + " custom tracks");
+        customTracks.forEach(e -> trackEntries.put(e.getName(), e));
+        this.tracks.addAll(loadByEntries(customTracks));
+
+
         //TODO use DB:
         List<String> trackPackagesNames = new ArrayList<>();
 
@@ -153,7 +179,7 @@ public final class TrackFactory {
             }
         }
 
-        allTracks.forEach(e -> trackEntries.put(e.getName(), e));
+
     }
 
     /**
@@ -187,7 +213,7 @@ public final class TrackFactory {
         exe.shutdown();
 
         try {
-            int timeout = (System.getenv("HOME").contains("menzel")) ? 30 : 60 * 2;
+            int timeout = (System.getenv("HOME").contains("menzel")) ? 60 : 60 * 2;
             completionService.poll(timeout, TimeUnit.SECONDS);
 
             logger.warn("Still loading track files. Stopping now");

backend/src/main/java/de/thm/run/AnalysisHelper.java

@@ -207,8 +207,10 @@ class AnalysisHelper {
         List<Sites> batchSites = command.getBatchSites();
         Sites bg;
 
+        int bgsitecount = Math.max(command.getMinBg(),batchSites.stream().map(s -> s.getPositions().size()).max(Integer::compareTo).get());
+
         if (command.getSitesBg() == null)
-            bg = BackgroundModelFactory.createBackgroundModel(command.getAssembly(), command.getMinBg());
+            bg = BackgroundModelFactory.createBackgroundModel(command.getAssembly(),bgsitecount);
         else
             bg = command.getSitesBg();
 

enhortlib/src/main/java/de/thm/misc/ChromosomSizes.java

@@ -47,12 +47,7 @@ public final class ChromosomSizes {
      */
     private ChromosomSizes() {
 
-        //TODO use db again?
-        if(System.getenv("HOME").contains("menzel")) {
-            chromosomeSizes = getChrSizesFromFile("/home/menzel/Desktop/THM/lfba/enhort/repo/backend/chrSizes.csv");
-        } else {
-            chromosomeSizes = getChrSizesFromFile("/home/mmnz21/enhort/chrSizes.csv");
-        }
+        chromosomeSizes = getChrSizes();
 
         for(Genome.Assembly assembly: chromosomeSizes.keySet()) {
             Map<String, Integer> hg = chromosomeSizes.get(assembly);
@@ -69,6 +64,68 @@ public final class ChromosomSizes {
         calcOffsets();
     }
 
+    private Map<Genome.Assembly, Map<String, Integer>> getChrSizes() {
+        String input = "hg19,chr1,249250621\n" +
+                "hg19,chr2,243199373\n" +
+                "hg19,chr3,198022430\n" +
+                "hg19,chr4,191154276\n" +
+                "hg19,chr5,180915260\n" +
+                "hg19,chr6,171115067\n" +
+                "hg19,chr7,159138663\n" +
+                "hg19,chr8,146364022\n" +
+                "hg19,chr9,141213431\n" +
+                "hg19,chr10,135534747\n" +
+                "hg19,chr11,135006516\n" +
+                "hg19,chr12,133851895\n" +
+                "hg19,chr13,115169878\n" +
+                "hg19,chr14,107349540\n" +
+                "hg19,chr15,102531392\n" +
+                "hg19,chr16,90354753\n" +
+                "hg19,chr17,81195210\n" +
+                "hg19,chr18,78077248\n" +
+                "hg19,chr19,59128983\n" +
+                "hg19,chr20,63025520\n" +
+                "hg19,chr21,48129895\n" +
+                "hg19,chr22,51304566\n" +
+                "hg19,chrX,155270560\n" +
+                "hg19,chrY,59373566\n" +
+                "GRCh38,chr1,248956422\n" +
+                "GRCh38,chr2,242193529\n" +
+                "GRCh38,chr3,198295559\n" +
+                "GRCh38,chr4,190214555\n" +
+                "GRCh38,chr5,181538259\n" +
+                "GRCh38,chr6,170805979\n" +
+                "GRCh38,chr7,159345973\n" +
+                "GRCh38,chr8,145138636\n" +
+                "GRCh38,chr9,138394717\n" +
+                "GRCh38,chr10,133797422\n" +
+                "GRCh38,chr11,135086622\n" +
+                "GRCh38,chr12,133275309\n" +
+                "GRCh38,chr13,114364328\n" +
+                "GRCh38,chr14,107043718\n" +
+                "GRCh38,chr15,101991189\n" +
+                "GRCh38,chr16,90338345\n" +
+                "GRCh38,chr17,83257441\n" +
+                "GRCh38,chr18,80373285\n" +
+                "GRCh38,chr19,58617616\n" +
+                "GRCh38,chr20,64444167\n" +
+                "GRCh38,chr21,46709983\n" +
+                "GRCh38,chr22,50818468\n" +
+                "GRCh38,chrX,156040895\n" +
+                "GRCh38,chrY,57227415";
+
+        Map<Genome.Assembly, Map<String, Integer>> chrs = new HashMap<>();
+
+        for( String line: input.split("\n")) {
+            String[] parts = line.split(",");
+
+            if (!chrs.containsKey(Genome.Assembly.valueOf(parts[0])))
+                chrs.put(Genome.Assembly.valueOf(parts[0]), new HashMap<>());
+            chrs.get(Genome.Assembly.valueOf(parts[0])).put(parts[1], Integer.parseInt(parts[2]));
+        }
+        return chrs;
+    }
+
     private Map<Genome.Assembly,Map<String,Integer>> getChrSizesFromFile(String filepath) {
         Map<Genome.Assembly, Map<String, Integer>> chrs = new HashMap<>();