Esempio n. 1
0
def runtest():
    failed = False
    sraob = sra.SRA('ERR726985', directory='./pyrpipe_sratest')
    if not sraob.fastq_exists():
        pu.print_boldred('Test failed')
        failed = True
    pu.print_notification('Cleaning up...')
    sraob.delete_fastq()
    os.rmdir(sraob.directory)

    if failed:
        pu.print_boldred('Paired end test failed')
        failed = False

    sraob = sra.SRA('SRR2134545', directory='./pyrpipe_sratest')
    if not sraob.fastq_exists():
        pu.print_boldred('Test failed')
        failed = True
    pu.print_notification('Cleaning up...')
    sraob.delete_fastq()
    os.rmdir(sraob.directory)

    if failed:
        pu.print_boldred('Single end test failed')
        failed = False

    if not failed:
        pu.print_green(
            '\n#####################All Tests Passed#####################\n')
        os.rmdir('./pyrpipe_sratest')
Esempio n. 2
0
def test_sra():
    srrID = testVars.srr
    print("Testing SRA...")
    testDir = testVars.testDir
    newOb = sra.SRA(srrID, testDir)
    assert newOb.srr_accession == srrID, "Failed SRA init"
    assert newOb.download_sra() == True, "Failed to download SRA file"
    assert newOb.sraFileExistsLocally(
    ) == True, "Failed to locate .sra file on disk"
    ##test fasterq-dump
    assert newOb.run_fasterqdump(delete_sra=True, **{
        "-f": "",
        "-t": testDir
    }) == True, "Failed FQ dump"
    assert newOb.fastqFilesExistsLocally(
    ) == True, "Failed to locate .fastq files on disk"
    assert newOb.sraFileExistsLocally(
    ) != True, "Failed to delete .sra files from disk"

    st = pyrpipe_session.save_session(filename="mySession",
                                      add_timestamp=False,
                                      out_dir=testVars.testDir)
    assert st == True, "Session save failed"
    st = pyrpipe_session.restore_session(testVars.testDir +
                                         "/mySession.pyrpipe")
    assert st == True, "Session restore failed"

    assert newOb.fastqFilesExistsLocally(
    ) == True, "Failed to locate .fastq files on disk after restore"
Esempio n. 3
0
def test_sra2():
    srrID=testVars.srr
    testDir=testVars.testDir
    newOb=sra.SRA(srrID,testDir+"/testsra")
    assert newOb.srr_accession==srrID,"Failed SRA init"
    assert newOb.fastq_exists()==True, "Failed to locate .fastq files on disk"
    assert newOb.sra_exists()==False, "SRA file exists. "
    #delete downloaded files
    assert newOb.delete_fastq()==True, "Failed to delete .fastq files from disk"
    
    
Esempio n. 4
0
def test_sra():
    srrID=testVars.srr
    print ("Testing SRA...")
    testDir=testVars.testDir
    newOb=sra.SRA(srrID,testDir+"/testsra")
    assert newOb.srr_accession==srrID,"Failed SRA init"
    assert newOb.fastq_exists()==True, "Failed to locate .fastq files on disk"
    assert newOb.sra_exists()==False, "Deleted SRA file exists. "
    
    assert newOb.download_sra()==True, "Failed to download SRA file"
    assert newOb.sra_exists()==True, "Failed to locate .sra file on disk"
    assert newOb.delete_sra()==True, "Failed to delete .sra files from disk"    
    assert newOb.sra_exists()==False, "Deleted SRA file exists. "
    #delete downloaded files
    assert newOb.delete_fastq()==True, "Failed to delete .fastq files from disk"
Esempio n. 5
0
def test_sra():
    srrID = testVars.srr
    print("Testing SRA...")
    testDir = testVars.testDir
    newOb = sra.SRA(srrID, testDir + "/testsra")
    assert newOb.srr_accession == srrID, "Failed SRA init"
    assert newOb.download_sra() == True, "Failed to download SRA file"
    assert newOb.sraFileExistsLocally(
    ) == True, "Failed to locate .sra file on disk"
    ##test fasterq-dump
    assert newOb.run_fasterqdump(delete_sra=True, **{
        "-f": "",
        "-t": testDir
    }) == True, "Failed FQ dump"
    assert newOb.fastqFilesExistsLocally(
    ) == True, "Failed to locate .fastq files on disk"
    assert newOb.sraFileExistsLocally(
    ) != True, "Failed to delete .sra files from disk"
    #delete downloaded files
    assert newOb.delete_fastq(
    ) == True, "Failed to delete .fastq files from disk"
Esempio n. 6
0
def test_sra():
    srrID = testVars.srr
    print("Testing SRA...")
    testDir = testVars.testDir
    newOb = sra.SRA(srrID, testDir)
    assert newOb.srr_accession == srrID, "Failed SRA init"
    assert newOb.fastq_exists(
    ) == True, "Failed to locate .fastq files on disk"
    assert newOb.sra_exists() == False, "Failed to delete .sra files from disk"

    st = pyrpipe_session.save_session(filename="mySession",
                                      add_timestamp=False,
                                      out_dir=testVars.testDir)
    assert st == True, "Session save failed"
    st = pyrpipe_session.restore_session(testVars.testDir +
                                         "/mySession.pyrpipe")
    assert st == True, "Session restore failed"

    assert newOb.fastq_exists(
    ) == True, "Failed to locate .fastq files on disk after restore"
    assert newOb.delete_fastq() == True, "failed to delete fastq"
Esempio n. 7
0
tr=assembly.Trinity()
#trOpts={"--seqType":"fq","--left":"/home/usingh/work/urmi/hoap/test/test_Trinity_Assembly/reads.left.fq.gz",
#        "--right":"/home/usingh/work/urmi/hoap/test/test_Trinity_Assembly/reads.right.fq.gz","--output":"/home/usingh/work/urmi/hoap/test/trinity_testout","--max_memory":"2G"}
#tr.run_trinity(verbose=True,**trOpts)

tr.perform_assembly(sraOb,verbose=True)
"""
"""
sr=sra.SRA(scan_path="/home/usingh/work/urmi/hoap/test/SRR5507495/")


# scan a dir to crteate SRA objects
#subdirs=next(os.walk(testDir))[1]

subdirs=[os.path.join(testDir, o) for o in os.listdir(testDir) 
                    if os.path.isdir(os.path.join(testDir,o))]

sraList=[]
for d in subdirs:
    try:
        thisSRA=sra.SRA(scan_path=d)
        sraList.append(thisSRA)
        print("FOUND data in "+d)
    except:
        print("no data in "+d)

"""
print("fstqTEST")
sob = sra.SRA('SRR5507495', testDir)
sob.download_fastq(procs=8)
Esempio n. 8
0
# -*- coding: utf-8 -*-
"""
Created on Tue Dec 10 19:06:34 2019

@author: usingh
"""

from pyrpipe import sra, mapping, assembly, qc, tools, pyrpipe_utils

workingDir = "/home/usingh/work/urmi/hoap/test"

athalRunsSmol = ['SRR1583780', 'SRR5507495', 'SRR5507442']
sraObjects = []

for x in athalRunsSmol:
    thisSraOb = sra.SRA(x, workingDir)
    if thisSraOb.downloadSRAFile():
        sraObjects.append(thisSraOb)
    else:
        print("Download failed:" + x)

print("Following runs downloaded:")
for ob in sraObjects:
    print(ob.srrAccession)

for ob in sraObjects:
    ob.runFasterQDump(deleteSRA=True,
                      **{
                          "-e": "8",
                          "-f": "",
                          "-t": workingDir
Esempio n. 9
0
def test_pipeline1():
    sraOb = sra.SRA(srr, workingDir)
    st = sraOb.download_sra()
    assert st == True, "SRA download failed"

    st = sraOb.run_fasterqdump(delete_sra=False,
                               **{
                                   "-e": "8",
                                   "-f": "",
                                   "-t": workingDir
                               })
    assert st == True, "fqdump failed"

    bbdOpts = {
        "ktrim": "r",
        "k": "23",
        "mink": "11",
        "qtrim": "'rl'",
        "trimq": "10",
        "--": ("-Xmx2g", ),
        "ref": testVars.bbdukAdapters
    }
    bbdOb = qc.BBmap(**bbdOpts)
    st = sraOb.perform_qc(bbdOb)
    assert st == True, "bbduk failed"

    tgOpts = {
        "--cores": "10",
        "-o": testVars.testDir,
        "--paired": "",
        "--": (fq1, fq2)
    }
    tg = qc.Trimgalore(**tgOpts)
    st = sraOb.perform_qc(tg)
    assert st == True, "tg failed"

    #runbowtie2
    bt = mapping.Bowtie2(bowtie2_index="")
    assert bt.check_index() == False, "Failed bowtie2 check_index"
    st = bt.build_index(testVars.testDir + "/btIndex", "bowtieIndex",
                        testVars.genome)
    assert st == True, "Failed to build bowtie2 index"
    st = bt.perform_alignment(sraOb)
    assert os.path.isfile(st) == True, "bowtie failed"

    hsOpts = {"--dta-cufflinks": "", "-p": "8"}
    hs = mapping.Hisat2(hisat2_index="", **hsOpts)
    st = hs.build_index(testVars.testDir, "hisatindex", testVars.genome)
    assert st == True, "Failed to build hisat2 index"
    #perform alignment with sraobject
    st = hs.perform_alignment(sraOb)
    assert os.path.isfile(st) == True, "hisat failed"

    hisatSam = st
    samOb = tools.Samtools(**{"-@": "8"})
    bam = samOb.sam_sorted_bam(hisatSam, delete_sam=False, delete_bam=False)
    assert os.path.isfile(bam) == True, "sam to bam failed"

    stie = assembly.Stringtie(reference_gtf=testVars.gtf)
    result = stie.perform_assembly(bam, out_dir=testVars.testDir)
    assert pu.check_files_exist(result) == True, "Failed stringtie"

    tr = assembly.Trinity()
    tr_out = tr.perform_assembly(sraOb, verbose=True)
    assert pu.check_files_exist(tr_out) == True, "Failed stringtie"

    kl = quant.Kallisto(kallisto_index="")
    assert kl.check_index() == False, "Failed kallisto check_index"
    st = kl.build_index(index_path=testVars.testDir + "/kallistoIndex",
                        index_name="kalIndex",
                        fasta=testVars.cdna)
    assert st == True, "Failed to build kallisto index"
    st = kl.perform_quant(sraOb)
    assert os.path.isdir(st) == True, "Failed to run kallisto"

    sl = quant.Salmon(salmon_index="")
    assert sl.check_index() == False, "Failed salmon check_index"
    st = sl.build_index(index_path=testVars.testDir + "/salmonIndex",
                        index_name="salIndex",
                        fasta=testVars.cdna)
    assert st == True, "Failed to build salmon index"

    st = sl.perform_quant(sraOb)
    assert os.path.isdir(st) == True, "Failed to run salmon"
Esempio n. 10
0
#new tests
# test samtools
sam=testDir+"/test_files/athaliana/mapping/hisat2.sam"
sm=tools.Samtools(threads=5)
bam1=sm.sam_to_bam(sam,out_suffix="test2",threads=3, delete_sam=False,verbose=True,quiet=False,logs=True,objectid="NA",**{"-@":"4"})
print(bam1)
bam2=sm.sort_bam(bam1,out_suffix="test2",threads=2,delete_bam=False,verbose=True,quiet=False,logs=True,objectid="NA")
print(bam2)
#bam=sm.sam_sorted_bam(sam,delete_sam=False,delete_bam=False)

txd=tools.Transdecoder()
infa="/Users/usingh/work/urmi/tests/txd/test.fa"
outdir=txd.run_transdecoder_longorfs(infa,out_dir="/Users/usingh/work/urmi/tests/txd/mtout1")
print(outdir)

poutdir="/Users/usingh/work/urmi/tests/txd/mypredout"
predout=txd.run_transdecoder_predict(infa,longorfs_dir=outdir,out_dir=poutdir)
print(predout)


newSRA=sra.SRA('SRR5507343',testDir)
newSRA.download_fastq()
#run trimgalore
tg=qc.Trimgalore(threads=4)  #specify to use 8 cores
bd=qc.BBmap(threads=4,max_memory=1)

newSRA.perform_qc(bd)
#newSRA.perform_qc(tg)


Esempio n. 11
0
                     genome=testVars.genome)
hsOpts = {"--dta-cufflinks": "", "-p": "8"}
hs = mapping.Hisat2(index=testVars.testDir + "/hisatindex",
                    genome=testVars.genome,
                    **hsOpts)
star = mapping.Star(index=os.path.join(testVars.testDir, "starIndex"),
                    genome=testVars.genome)
samOb = tools.Samtools()
stie = assembly.Stringtie()
kl = quant.Kallisto(index=testVars.testDir + "/kallistoIndex/kalIndex",
                    transcriptome=testVars.cdna)
sl = quant.Salmon(index=testVars.testDir + "/salmonIndex/salIndex",
                  transcriptome=testVars.cdna_big)

#sra ob
sraOb = sra.SRA(srr, workingDir)
st = sraOb.fastq_exists()
assert st == True, "fasterq-dump failed"


def test_pipeline1():
    st = sraOb.trim(bbdOb).align(hs).assemble(stie).quant(kl)
    assert st != None, "pipeline 1 failed"


def test_pipeline2():
    st = sraOb.trim(tg).align(hs).assemble(stie).quant(kl)
    assert st != None, "pipeline 2 failed"


def test_pipeline3():
Esempio n. 12
0
        progresslog.write(error + '\n')
        continue

    #convert bam to fastq
    st = biobb.bamtofq(bam, sampleid)
    if not st:
        print('bam to fastq failed.')
        #log this message
        error = sampleid + '\tbam to fastq failed'
        progresslog.write(error + '\n')
        continue

    #create SRA object
    fq1 = os.path.join(bam_dir, sampleid + '_1.fastq')
    fq2 = os.path.join(bam_dir, sampleid + '_2.fastq')
    sobj = sra.SRA(fastq=fq1, fastq2=fq2)
    #check files exist
    if not _dryrun and not sobj.fastq_exists():
        print('Error creating SRA obj')
        error = sampleid + '\tFastq not found'
        progresslog.write(error + '\n')
        continue

    if runalign:
        #perform alignment
        #start star alignment
        star_bam = starobj.perform_alignment(sobj)  #returns bam path
        #delete fastq
        sobj.delete_fastq()
        if not star_bam:
            print('Star failed for ' + bam)
Esempio n. 13
0
#check if hisat is sucessful
if not hisatSam:
    raise Exception("ERROR: Hisat failed")
    
#remove qc corrected fastq
sraOb.deleteFastqFiles()
#run sam to sorted bam then run stringtie
gtfS=stieOb.runStringtie(samtOb.samToSortedBam(hisatSam,10,deleteSam=True,deleteOriginalBam=True),deleteInputBam=True,proc=10)
"""



btIndex="/home/usingh/work/urmi/hoap/test/bowtieIndex/rRNAindex"
#riboseq SRR3590744
sraOb=sra.SRA(srr_accession='SRR5507495',location=testDir)
#download sra
sraOb.download_sra()
#run fastqdump;delete sra when done
sraOb.run_fasterqdump(delete_sra=False,**{"-f":"","-t":testDir})

tgOb=qc.Trimgalore()

sraOb.perform_qc(tgOb)

#pathToAdapters="/home/usingh/lib_urmi/softwares/bbmap/resources/adapters2.fa"
#bbdOpts={"ktrim":"r","k":"23","mink":"11","qtrim":"'rl'","trimq":"10","--":("-Xmx2g",),"ref":pathToAdapters}
#bbdOb=qc.BBmap(**bbdOpts)

#sraOb.perform_qc(bbdOb)
#status=bbdOb.performCleaning(sraOb,"/home/usingh/work/urmi/hoap/test/bowtieIndex/euk_combined_rRNA.fa")