Option to create readsets from SRA (#355)

* Option to create readsets from SRA
* Test fix

Author: bbimber

SequenceAnalysis/resources/queries/sequenceanalysis/sequence_readsets.js

@@ -1,11 +1,15 @@
-/*
- * Copyright (c) 2012 LabKey Corporation
- *
- * Licensed under the Apache License, Version 2.0: http://www.apache.org/licenses/LICENSE-2.0
- */
+var LABKEY = require("labkey");
+
+var triggerHelper = new org.labkey.sequenceanalysis.query.SequenceTriggerHelper(LABKEY.Security.currentUser.id, LABKEY.Security.currentContainer.id);
 
 function beforeDelete(row, errors){
     if (!this.extraContext.deleteFromServer){
         errors._form = 'You cannot directly delete readsets.  To delete these records, use the delete button above the readset grid.';
     }
+}
+
+function afterInsert(row, errors) {
+    if (row.sraAccessions) {
+        triggerHelper.createReaddataForSra(row.rowid, row.sraAccessions);
+    }
 }

SequenceAnalysis/src/org/labkey/sequenceanalysis/query/SequenceTriggerHelper.java

@@ -13,9 +13,11 @@
 import org.biojava.nbio.core.sequence.transcription.TranscriptionEngine;
 import org.junit.Assert;
 import org.junit.Test;
+import org.labkey.api.assay.AssayFileWriter;
 import org.labkey.api.data.Container;
 import org.labkey.api.data.ContainerManager;
 import org.labkey.api.data.SimpleFilter;
+import org.labkey.api.data.Table;
 import org.labkey.api.data.TableInfo;
 import org.labkey.api.data.TableSelector;
 import org.labkey.api.exp.api.DataType;
@@ -28,12 +30,19 @@
 import org.labkey.api.security.User;
 import org.labkey.api.security.UserManager;
 import org.labkey.api.sequenceanalysis.RefNtSequenceModel;
+import org.labkey.api.util.FileUtil;
+import org.labkey.api.util.Path;
+import org.labkey.sequenceanalysis.ReadDataImpl;
 import org.labkey.sequenceanalysis.SequenceAnalysisSchema;
+import org.labkey.sequenceanalysis.SequenceAnalysisServiceImpl;
+import org.labkey.sequenceanalysis.SequenceReadsetImpl;
+import org.labkey.vfs.FileLike;
 
 import java.io.File;
 import java.io.IOException;
 import java.io.StringWriter;
 import java.util.Arrays;
+import java.util.Date;
 import java.util.HashMap;
 import java.util.List;
 import java.util.Map;
@@ -254,4 +263,64 @@ public int createExpData(String relPath) {
 
         return d.getRowId();
     }
+
+    public void createReaddataForSra(int readsetId, String sraAccessions)
+    {
+        SequenceReadsetImpl rs = SequenceAnalysisServiceImpl.get().getReadset(readsetId, _user);
+        if (rs == null)
+        {
+            throw new IllegalArgumentException("Unable to find readset: " + readsetId);
+        }
+
+        TableInfo rd = SequenceAnalysisSchema.getTable(SequenceAnalysisSchema.TABLE_READ_DATA);
+
+        String[] tokens = StringUtils.split(sraAccessions, ",");
+        for (String token : tokens)
+        {
+            if (rs.getReadData() != null && !rs.getReadData().isEmpty())
+            {
+                throw new IllegalArgumentException("Did not expect readset to have existing readdata: " + rs.getReadsetId());
+            }
+
+            // Create new:
+            ReadDataImpl rd1 = new ReadDataImpl();
+            rd1.setReadset(readsetId);
+            rd1.setContainer(rs.getContainer());
+            rd1.setCreated(new Date());
+            rd1.setModified(new Date());
+            rd1.setCreatedBy(_user.getUserId());
+            rd1.setModifiedBy(_user.getUserId());
+            rd1.setSra_accession(token);
+            rd1.setArchived(true);
+
+            // NOTE: this is a fragile assumption. We might need to eventually query SRA to figure out whether data is paired:
+            Container c = ContainerManager.getForId(rs.getContainer());
+            PipeRoot pr = PipelineService.get().findPipelineRoot(c);
+            if (pr == null)
+            {
+                throw new IllegalStateException("Unable to find pipeline root for: " + c.getPath());
+            }
+
+            String folderName = "SequenceImport_" + FileUtil.getTimestamp();
+            FileLike outDir = AssayFileWriter.findUniqueFileName(folderName, pr.getRootFileLike());
+
+            FileLike expectedFile1 = FileUtil.appendPath(outDir, Path.parse(token + "_1.fastq.gz"));
+            ExpData exp1 = ExperimentService.get().createData(c, new DataType("Data"));
+            exp1.setDataFileURI(expectedFile1.toURI());
+            exp1.setContainer(c);
+            exp1.setName(expectedFile1.getName());
+            exp1.save(_user);
+            rd1.setFileId1(exp1.getRowId());
+
+            FileLike expectedFile2 = FileUtil.appendPath(outDir, Path.parse(token + "_2.fastq.gz"));
+            ExpData exp2 = ExperimentService.get().createData(c, new DataType("Data"));
+            exp2.setDataFileURI(expectedFile2.toURI());
+            exp2.setContainer(c);
+            exp2.setName(expectedFile2.getName());
+            exp2.save(_user);
+            rd1.setFileId2(exp2.getRowId());
+
+            Table.insert(_user, rd, rd1);
+        }
+    }
 }

SequenceAnalysis/test/src/org/labkey/test/tests/external/labModules/SequenceTest.java

@@ -195,6 +195,44 @@ private void importReadsetMetadata()
         log("verifying readset count correct");
         waitForText("Sequence Readsets");
         waitForElement(LabModuleHelper.getNavPanelItem("Sequence Readsets:", _readsetCt.toString()));
+
+        // Repeat, adding SRA accessions:
+        goToProjectHome();
+        waitAndClick(Locator.linkWithText("Plan Sequence Run (Create Readsets)"));
+        new Window.WindowFinder(getDriver()).withTitle("Create Readsets").waitFor();
+        waitAndClickAndWait(Ext4Helper.Locators.ext4Button("Submit"));
+
+        _helper.waitForField("Sample Id", WAIT_FOR_PAGE);
+        _ext4Helper.clickTabContainingText("Import Spreadsheet");
+        waitForText("Copy/Paste Data");
+
+        setFormElementJS(Locator.name("text"), getIlluminaSRANames());
+
+        waitAndClick(Ext4Helper.Locators.ext4Button("Upload"));
+        new Window.WindowFinder(getDriver()).withTitle("Success").waitFor();
+        _readsetCt += 3;
+        assertTextPresent("Success!");
+        waitAndClickAndWait(Ext4Helper.Locators.ext4Button("OK"));
+
+        // This is scoped to this workbook:
+        log("verifying readset count correct");
+        waitForText("Sequence Readsets");
+        waitAndClickAndWait(LabModuleHelper.getNavPanelItem("Sequence Readsets:", "3"));
+
+        DataRegionTable.DataRegion(getDriver()).withName("query").waitFor();
+
+        //verify CSV file creation
+        DataRegionTable.DataRegion(getDriver()).find().goToView("SRA Info");
+        DataRegionTable dr = DataRegionTable.DataRegion(getDriver()).withName("query").waitFor();
+        waitForElement(Locator.tagContainingText("a", "SRA0"));
+        waitForElement(Locator.tagContainingText("a", "SRA1"));
+        waitForElement(Locator.tagContainingText("a", "SRA2"));
+
+        dr.checkAllOnPage();
+        dr.clickHeaderButtonAndWait("Delete");
+        clickButton("OK");
+
+        _readsetCt -= 3;
     }
 
     /**
@@ -320,6 +358,18 @@ private String getIlluminaNames()
         return sb.toString();
     }
 
+    private String getIlluminaSRANames()
+    {
+        StringBuilder sb = new StringBuilder("Name\tPlatform\tsraAccessions\n");
+        int i = 0;
+        while (i < 3)
+        {
+            sb.append("IlluminaSRA" + (i + 1) + "\tILLUMINA\tSRA" + i + "\n");
+            i++;
+        }
+        return sb.toString();
+    }
+
     /**
      * This test will kick off a pipeline import using the illumina pipeline.  Verification of the result
      * is performed by readsetFeaturesTest()