annotate commons/launcher/tests/Test_F_LaunchTallymer.py @ 18:94ab73e8a190

Uploaded

import unittest
import os
from commons.core.utils.FileUtils import FileUtils
import shutil
from commons.launcher.LaunchTallymer import LaunchTallymer

class Test_F_LaunchTallymer(unittest.TestCase):

    def setUp(self):
        self._inFastaFileName = "%s/Tools/DmelChr4.fa" % os.environ["REPET_DATA"]

    def test_run_output_as_gff_cleaned(self):
        iLaunchTallymer = LaunchTallymer(self._inFastaFileName,  clean = True, verbosity=2)
        iLaunchTallymer.run()
        
        expFileName = "%s/Tools/DmelChr4.fa.tallymer_k20.gff" % os.environ["REPET_DATA"]
        obsFileName = "DmelChr4.tallymer.gff"
        
        self.assertTrue(FileUtils.are2FilesIdentical(expFileName, obsFileName))
        os.remove(obsFileName)
        os.remove("DmelChr4.tallymer.stat")
        
    def test_run_with_remote_dir(self):
        os.makedirs("sourceDir")
        changedInFastaFileName = os.path.basename(self._inFastaFileName)
        changedInFastaFilepath = os.path.join("sourceDir",changedInFastaFileName)
        shutil.copy2(self._inFastaFileName, changedInFastaFilepath)
        
        iLaunchTallymer = LaunchTallymer(changedInFastaFilepath,  clean = True, verbosity=2)
        iLaunchTallymer.run()
        
        expFileName = "%s/Tools/DmelChr4.fa.tallymer_k20.gff" % os.environ["REPET_DATA"]
        obsFileName = "DmelChr4.tallymer.gff"
        
        
        self.assertTrue(FileUtils.are2FilesIdentical(expFileName, obsFileName))
        obsSourceDirFiles = os.listdir("sourceDir")
        expSourceDirFiles = [changedInFastaFileName,]
        self.assertEquals(expSourceDirFiles, obsSourceDirFiles)
        
        os.remove(obsFileName)
        os.remove("DmelChr4.tallymer.stat")
        shutil.rmtree("sourceDir")


    def test_run_output_as_gff_cleaned_changed_minOccs2(self):
        iLaunchTallymer = LaunchTallymer(self._inFastaFileName, minOccs=2, clean = True, verbosity=2)
        iLaunchTallymer.run()
        
        expFileName = "%s/Tools/DmelChr4.fa.tallymer_k20_minOcc2.gff" % os.environ["REPET_DATA"]
        obsFileName = "DmelChr4.tallymer.gff"
        
        self.assertTrue(FileUtils.are2FilesIdentical(expFileName, obsFileName))
        os.remove(obsFileName)
        os.remove("DmelChr4.tallymer.stat")
        
    def test_run_output_as_gff_cleaned_relative_input(self):
        srcFastaFileName = self._inFastaFileName
        dstFastaFileName = os.path.basename(self._inFastaFileName)
        shutil.copy2(srcFastaFileName, dstFastaFileName)
        iLaunchTallymer = LaunchTallymer(dstFastaFileName,  clean = True, verbosity=2)
        iLaunchTallymer.run()
        
        expFileName = "%s/Tools/DmelChr4.fa.tallymer_k20.gff" % os.environ["REPET_DATA"]
        obsFileName = "DmelChr4.tallymer.gff"
        
        self.assertTrue(FileUtils.are2FilesIdentical(expFileName, obsFileName))
        
        os.remove(obsFileName)
        os.remove("DmelChr4.tallymer.stat")
        os.remove(dstFastaFileName)
   
    def test_run_as_script_output_as_gff_cleaned_relative_input_(self):
        srcFastaFileName = self._inFastaFileName
        dstFastaFileName = os.path.basename(self._inFastaFileName)
        shutil.copy2(srcFastaFileName, dstFastaFileName)
        
        cmd = 'LaunchTallymer.py -i %s -s 20 -c -v 3' % (self._inFastaFileName)
        os.system(cmd)
        expFileName = "%s/Tools/DmelChr4.fa.tallymer_k20.gff" % os.environ["REPET_DATA"]
        obsFileName = "DmelChr4.tallymer.gff"
        
        self.assertTrue(FileUtils.are2FilesIdentical(expFileName, obsFileName))
        
        os.remove(obsFileName)
        os.remove("DmelChr4.tallymer.stat")
        os.remove(dstFastaFileName)
        
    def test_run_output_as_wig_cleaned(self):
        iLaunchTallymer = LaunchTallymer(self._inFastaFileName, outputFormats="wig", clean=True)
        iLaunchTallymer.run()
        
        expFileName = "%s/Tools/DmelChr4.fa.tallymer_k20.wig" % os.environ["REPET_DATA"]
        obsFileName = "DmelChr4.tallymer.wig"
        
        self.assertTrue(FileUtils.are2FilesIdentical(expFileName, obsFileName))
        
        os.remove(obsFileName)
        os.remove("DmelChr4.tallymer.stat")

    def test_run_output_as_gff_and_wig_k15_cleaned(self):
        iLaunchTallymer = LaunchTallymer(self._inFastaFileName, merSize=15, outputFormats="wig,gff3", clean=True)
        iLaunchTallymer.run()
        
        expFileName = "%s/Tools/DmelChr4.fa.tallymer_k15.wig" % os.environ["REPET_DATA"]
        obsFileName = "DmelChr4.tallymer.wig"
        self.assertTrue(FileUtils.are2FilesIdentical(expFileName, obsFileName))
        os.remove(obsFileName)
        
        expFileName = "%s/Tools/DmelChr4.fa.tallymer_k15.gff3" % os.environ["REPET_DATA"]
        obsFileName = "DmelChr4.tallymer.gff3"
        self.assertTrue(FileUtils.are2FilesIdentical(expFileName, obsFileName))
        os.remove(obsFileName)
        
        os.remove("DmelChr4.tallymer.stat")

    def test_run_as_script_run_output_as_gff_and_wig_k15_cleaned(self):
        logFileName = "test.log"
        cmd = 'LaunchTallymer.py -i %s -s 15 -f gff,wig,vero -c -v 3 > %s' % (self._inFastaFileName, logFileName)
        os.system(cmd)
               
        expFileName = "%s/Tools/DmelChr4.fa.tallymer_k15.wig" % os.environ["REPET_DATA"]
        obsFileName = "DmelChr4.tallymer.wig"
        self.assertTrue(FileUtils.are2FilesIdentical(expFileName, obsFileName))
        os.remove(obsFileName)
        
        expFileName = "%s/Tools/DmelChr4.fa.tallymer_k15.gff3" % os.environ["REPET_DATA"]
        obsFileName = "DmelChr4.tallymer.gff"
        self.assertTrue(FileUtils.are2FilesIdentical(expFileName, obsFileName))
        os.remove(obsFileName)
        
        expLogFileName = "exp.log"
        with open(expLogFileName, "w") as f:
            f.write(" LaunchTallymer - WARNING - Warning: ignoring invalid formats: <vero>\n")
            f.write(" LaunchTallymer - INFO - Starting to run tallymer search of sequence DmelChr4.fa \n")
            f.write(" LaunchTallymer - INFO - Finished running tallymer scan of sequence DmelChr4.fa \n")
            f.write(" LaunchTallymer - INFO - Starting to run tallymer search to map conversion\n")
            f.write(" LaunchTallymer - INFO - Finished tallymer search to map conversion\n")
            f.write(" LaunchTallymer - INFO - Generating wig file\n")
            f.write(" LaunchTallymer - INFO - Generating gff file\n")
        obsLogFileName = "obs.log"
        os.system("cut -d'-' -f4,5,6 %s > %s" % (logFileName, obsLogFileName))
        self.assertTrue(FileUtils.are2FilesIdentical(expLogFileName, obsLogFileName))
        
        os.remove(logFileName)
        os.remove(expLogFileName)
        os.remove(obsLogFileName)
        os.remove("DmelChr4.tallymer.stat")
        
if __name__ == "__main__":
    unittest.main()