Python ngs.NGS类代码示例

def run(acc, splitNum, splitNo):
    # open requested accession using SRA implementation of the API
    with NGS.openReadCollection(acc) as run:
        run_name = run.getName()
        
        # compute window to iterate through
        MAX_ROW = run.getAlignmentCount()
        chunk = MAX_ROW / splitNum
        first = int(round(chunk * (splitNo-1)))
        next_first = int(round(chunk * (splitNo)))
        if next_first > MAX_ROW:
            next_first = MAX_ROW

        # start iterator on reads
        with run.getAlignmentRange(first+1, next_first-first, Alignment.primaryAlignment) as it:
            i = 0
            while it.nextAlignment():
                print ("{}\t{}\t{}\t{}\t{}\t{}".format(it.getReadId(),
                    it.getReferenceSpec(),
                    it.getAlignmentPosition(),
                    it.getShortCigar(False),
                    it.getFragmentBases(),
                    ("aligned" if it.isAligned() else "unaligned"),
                ))
                i += 1
            print ("Read {} alignments for {}".format(i, run_name))

 def test_ReadGroupIterator_ThrowBeforeNext(self):
     it = NGS.openReadCollection(PrimaryOnly).getReadGroups()
     try:
         it.getName()
         self.fail()
     except ErrorMsg:
         pass

def run(acc, splitNum, splitNo):
    # open requested accession using SRA implementation of the API
    with NGS.openReadCollection(acc) as run:
        run_name = run.getName()

        # compute window to iterate through
        MAX_ROW = run.getReadCount()
        chunk = MAX_ROW/splitNum
        first = int(round(chunk*(splitNo-1)))
        next_first = int(round(chunk*(splitNo)))
        if next_first > MAX_ROW:
            next_first = MAX_ROW

        # start iterator on reads
        with run.getReadRange(first+1, next_first-first, Read.all) as it:
            i = 0
            while it.nextRead():
                i += 1
                print (it.getReadId())
                # iterate through fragments
                while it.nextFragment():
                    bases = it.getFragmentBases()
                    if bases:
                        print ("\t{} - {}".format(bases, "aligned" if it.isAligned() else "unaligned"))
                print ("\n")
            print ("Read {} spots for {}".format(i,  run_name))

def run(acc, splitNum, splitNo):  # this function doesn't release NGS objects however it might
    # open requested accession using SRA implementation of the API
    run = NGS.openReadCollection(acc)
    run_name = run.getName()

    # compute window to iterate through
    MAX_ROW = run.getAlignmentCount()
    chunk = MAX_ROW / splitNum
    first = int(round(chunk * (splitNo - 1)))
    next_first = int(round(chunk * (splitNo)))
    if next_first > MAX_ROW:
        next_first = MAX_ROW

    # start iterator on reads
    it = run.getAlignmentRange(first + 1, next_first - first, Alignment.primaryAlignment)
    i = 0
    while it.nextAlignment():
        print(
            it.getReadId()
            + "\t"
            + it.getReferenceSpec()
            + "\t"
            + str(it.getAlignmentPosition())
            + "\t"
            + it.getShortCigar(False)
            + "\t"
            + it.getFragmentBases()
            + "\t"
            + ("aligned" if it.isAligned() else "unaligned")
        )
        i += 1
    print("Read {} alignments for {}".format(i, run_name))

def run(acc, splitNum, splitNo): # this function doesn't release NGS objects however it might
    # open requested accession using SRA implementation of the API
    run = NGS.openReadCollection(acc)
    run_name = run.getName()

    # compute window to iterate through
    MAX_ROW = run.getReadCount()
    chunk = MAX_ROW/splitNum
    first = int(round(chunk*(splitNo-1)))
    next_first = int(round(chunk*(splitNo)))
    if next_first > MAX_ROW:
        next_first = MAX_ROW

    # start iterator on reads
    it = run.getReadRange(first+1, next_first-first, Read.all)
    i = 0
    while it.nextRead():
        i += 1
        print (it.getReadId())
        # iterate through fragments
        while it.nextFragment():
            bases = it.getFragmentBases()
            if bases:
                print ("\t" + bases + " - " + ("aligned" if it.isAligned() else "unaligned"))
        print ("\n")
    print ("Read {} spots for {}".format(i,  run_name))

def run(acc, refName, start, stop):
    # open requested accession using SRA implementation of the API
    with NGS.openReadCollection(acc) as run:
        run_name = run.getName()

        # get requested reference
        with run.getReference(refName) as ref:
            # start iterator on requested range
            with ref.getPileupSlice(start - 1, stop - start + 1) as it:
                i = 0
                while it.nextPileup():
                    qual = ""
                    base = ""

                    line = "{}\t{}\t{}\t{}".format(
                        it.getReferenceSpec(), it.getReferencePosition() + 1, it.getReferenceBase(), it.getPileupDepth()
                    )
                    while it.nextPileupEvent():
                        e = it.getEventType()
                        if (e & PileupEvent.alignment_start) != 0:
                            base = base + "^"
                            base = base + chr(it.getMappingQuality() + 33)

                        if (e & PileupEvent.insertion) != 0:
                            base = base + "+"
                            ibases = it.getInsertionBases()
                            c = len(ibases)
                            base = base + str(c)

                            if (e & PileupEvent.alignment_minus_strand) != 0:
                                base = base + ibases.lower()
                            else:
                                base = base + ibases

                        evt = e & 7

                        if (e & PileupEvent.alignment_minus_strand) != 0:
                            if evt == PileupEvent.deletion:
                                base = base + "<"
                            elif evt == PileupEvent.match:
                                base = base + ","
                            elif evt == PileupEvent.mismatch:
                                base = base + str(it.getAlignmentBase()).lower()
                        else:
                            if evt == PileupEvent.deletion:
                                base = base + ">"
                            elif evt == PileupEvent.match:
                                base = base + "."
                            elif evt == PileupEvent.mismatch:
                                base = base + str(it.getAlignmentBase()).upper()

                        if (e & PileupEvent.alignment_stop) != 0:
                            base = base + "$"

                        qual = qual + it.getAlignmentQuality()

                    i += 1
                    print("{}\t{}\t{}".format(line, base, qual))
                print("Read {} pileups for {}".format(i, run_name))

 def test_ReferenceWindow_Slice_Filtered_Category (self):
     it = NGS.openReadCollection(WithSecondary).getReference("gi|169794206|ref|NC_010410.1|").getAlignmentSlice(516000, 100000, Alignment.primaryAlignment) 
     self.assertTrue(it.nextAlignment())  
     self.assertEqual(WithSecondary + ".PA.33", it. getAlignmentId())
     self.assertTrue(it.nextAlignment())  
     self.assertEqual(WithSecondary + ".PA.34", it. getAlignmentId())
     self.assertTrue(it.nextAlignment())  
     self.assertEqual(WithSecondary + ".PA.35", it. getAlignmentId()) # no secondary
     self.assertFalse(it.nextAlignment())

    def test_ReadGroup_getStatistics(self):
        gr = NGS.openReadCollection(WithGroups).getReadGroup("GS57510-FS3-L03")

        stats = gr.getStatistics()
    
        self.assertEqual(34164461870, stats.getAsU64("BASE_COUNT"))
        self.assertEqual(34164461870, stats.getAsU64("BIO_BASE_COUNT"))
        self.assertEqual(488063741,   stats.getAsU64("SPOT_COUNT"))
        self.assertEqual(5368875807,  stats.getAsU64("SPOT_MAX"))
        self.assertEqual(4880812067,  stats.getAsU64("SPOT_MIN"))

 def test_Alignment_isPaired_MultiFragmentsPerSpot(self):
     readCollection = NGS.openReadCollection(PrimaryOnly)
     alignment = readCollection.getAlignment(PrimaryOnly + ".PA.1")
     self.assertTrue(alignment.isPaired())
     
     alignment = readCollection.getAlignment(PrimaryOnly + ".PA.2")
     self.assertTrue(alignment.isPaired())
     
     # has unaligned mate
     alignment = readCollection.getAlignment (PrimaryOnly + ".PA.6")
     self.assertTrue(alignment.isPaired())

 def test_ReferenceWindow_Slice_Filtered_Start_Within_Slice (self):
     ref = NGS.openReadCollection(WithCircularRef).getReference("NC_012920.1")
     it = ref.getFilteredAlignmentSlice(0, ref.getLength(), Alignment.all, Alignment.startWithinSlice, 0)
 
     self.assertTrue(it.nextAlignment())
     lastAlignmentPosition = it.getAlignmentPosition()
     while it.nextAlignment():
         currentPosition = it.getAlignmentPosition()
         errorMsg = "Sorting violated. Last position (" + str(lastAlignmentPosition) + ") is higher than current one (" + str(currentPosition) + ")"
         self.assertTrue ( lastAlignmentPosition <= currentPosition, errorMsg )
         lastAlignmentPosition = currentPosition

def run(acc, refName=None):
    # open requested accession using SRA implementation of the API
    with NGS.openReadCollection(acc) as run:
        if refName:
            with run.getReference(refName) as ref:
                process(ref)
        else:
            with run.getReferences() as refs:
                while refs.nextReference():
                    process(refs)
                    print("")

def run(acc):
    # open requested accession using SRA implementation of the API
    with NGS.openReadCollection(acc) as run:
        run_name = run.getName()

        # get requested reference
        with run.getReferences() as it:
            i = 0
            while it.nextReference():
                print ("{}\t{}\t{}\t{}".format(it.getCommonName(),
                    it.getCanonicalName(),
                    it.getLength(),
                    "circular" if it.getIsCircular() else "linear",
                ))

            print ("Read {} references for {}".format(i, run_name))

def run(acc): # this function doesn't release NGS objects however it might
    # open requested accession using SRA implementation of the API
    run = NGS.openReadCollection(acc)
    run_name = run.getName()

    # get requested reference
    it = run.getReferences()

    i = 0
    while it.nextReference():
        print (
            it.getCommonName() + "\t" +
            it.getCanonicalName() + "\t" +
            str(it.getLength()) + "\t" +
            ("circular" if it.getIsCircular() else "linear")
        )
    print ("Read {} references for {}".format(i, run_name))

    def test_ReferenceWindow(self):
        it = NGS.openReadCollection(WithSecondary).getReference("gi|169794206|ref|NC_010410.1|").getAlignments(Alignment.all) 
        self.assertTrue(it.nextAlignment())  
    
        # the first 2 secondary alignments' locations on the list: #34, #61
        count = 1;
        while it.nextAlignment():
            if it.getAlignmentCategory() == Alignment.secondaryAlignment:
                break
            count += 1

        self.assertEqual(34, count)    
        while it.nextAlignment():
            if it.getAlignmentCategory() == Alignment.secondaryAlignment:
                break
            count += 1

        self.assertEqual(61, count)

def run(acc, refName, start, stop):
    # open requested accession using SRA implementation of the API
    with NGS.openReadCollection(acc) as run:
        run_name = run.getName()
    
        # get requested reference
        with run.getReference(refName) as ref:
            # start iterator on requested range
            with ref.getAlignmentSlice(start, stop-start+1, Alignment.primaryAlignment) as it:
                i = 0
                while it.nextAlignment():
                    print ("{}\t{}\t{}\t{}\t{}".format(
                        it.getReadId(),
                        it.getReferenceSpec(),
                        it.getAlignmentPosition(),
                        it.getLongCigar(False),
                        it.getAlignedFragmentBases(),
                        ))
                    i += 1
                print ("Read {} alignments for {}".format(i, run_name))