dexseq_count.py

import sys, itertools, optparse, warnings

optParser = optparse.OptionParser(

   usage = "python %prog [options] <flattened_gff_file> <alignment_file> <output_file>",

   description=
      "This script counts how many reads in <alignment_file> fall onto each exonic " +
      "part given in <flattened_gff_file> and outputs a list of counts in " +
      "<output_file>, for further analysis with the DEXSeq Bioconductor package. " +
      "Notes: Use dexseq_prepare_annotation.py to produce <flattened_gff_file>. " +
      "<alignment_file> may be '-' to indicate standard input.",

   epilog =
      "Written by Simon Anders (sanders@fs.tum.de) and Alejandro Reyes (reyes@embl.de), " +
      "European Molecular Biology Laboratory (EMBL). (c) 2010-2013. Released under the " +
      " terms of the GNU General Public License v3. Part of the 'DEXSeq' package." )

optParser.add_option( "-p", "--paired", type="choice", dest="paired",
   choices = ( "no", "yes" ), default = "no",
   help = "'yes' or 'no'. Indicates whether the data is paired-end (default: no)" )

optParser.add_option( "-s", "--stranded", type="choice", dest="stranded",
   choices = ( "yes", "no", "reverse" ), default = "yes",
   help = "'yes', 'no', or 'reverse'. Indicates whether the data is " +
      "from a strand-specific assay (default: yes ). " +
      "Be sure to switch to 'no' if you use a non strand-specific RNA-Seq library " +
      "preparation protocol. 'reverse' inverts strands and is needed for certain " +
      "protocols, e.g. paired-end with circularization."  )

optParser.add_option( "-a", "--minaqual", type="int", dest="minaqual",
   default = 10,
   help = "skip all reads with alignment quality lower than the given " +
      "minimum value (default: 10)" )

optParser.add_option( "-f", "--format", type="choice", dest="alignment",
   choices=("sam", "bam"), default="sam",
   help = "'sam' or 'bam'. Format of <alignment file> (default: sam)" )

optParser.add_option( "-r", "--order", type="choice", dest="order",
   choices=("pos", "name"), default="name",
   help = "'pos' or 'name'. Sorting order of <alignment_file> (default: name). Paired-end sequencing " +
      "data must be sorted either by position or by read name, and the sorting order " +
      "must be specified. Ignored for single-end data." )


if len( sys.argv ) == 1:
   optParser.print_help()
   sys.exit(1)

(opts, args) = optParser.parse_args()

if len( args ) != 3:
   sys.stderr.write( sys.argv[0] + ": Error: Please provide three arguments.\n" )
   sys.stderr.write( "  Call with '-h' to get usage information.\n" )
   sys.exit( 1 )

try:
   import HTSeq
except ImportError:
   sys.stderr.write( "Could not import HTSeq. Please install the HTSeq Python framework\n" )
   sys.stderr.write( "available from http://www-huber.embl.de/users/anders/HTSeq\n" )
   sys.exit(1)

gff_file = args[0]
sam_file = args[1]
out_file = args[2]
stranded = opts.stranded == "yes" or opts.stranded == "reverse"
reverse = opts.stranded == "reverse"
is_PE = opts.paired == "yes"
alignment = opts.alignment
minaqual = opts.minaqual
order = opts.order

if alignment == "bam":
   try:
      import pysam
   except ImportError:
      sys.stderr.write( "Could not import pysam, which is needed to process BAM file (though\n" )
      sys.stderr.write( "not to process text SAM files). Please install the 'pysam' library from\n" )
      sys.stderr.write( "https://code.google.com/p/pysam/\n" )
      sys.exit(1)



if sam_file == "-":
   sam_file = sys.stdin


# Step 1: Read in the GFF file as generated by aggregate_genes.py
# and put everything into a GenomicArrayOfSets

features = HTSeq.GenomicArrayOfSets( "auto", stranded=stranded )
for f in  HTSeq.GFF_Reader( gff_file ):
   if f.type == "exonic_part":
      f.name = f.attr['gene_id'] + ":" + f.attr['exonic_part_number']
      features[f.iv] += f

# initialise counters
num_reads = 0
counts = {}
counts[ '_empty' ] = 0
counts[ '_ambiguous' ] = 0
counts[ '_lowaqual' ] = 0
counts[ '_notaligned' ] = 0
counts['_ambiguous_readpair_position'] = 0

# put a zero for each feature ID
for iv, s in features.steps():
   for f in s:
      counts[ f.name ] = 0

#We need this little helper below:
def reverse_strand( s ):
   if s == "+":
      return "-"
   elif s == "-":
      return "+"
   else:
      raise SystemError, "illegal strand"

def update_count_vector( counts, rs ):
   if( type(rs) == str):
      counts[ rs ] += 1
   else:
      for f in rs:
         counts[f.name] += 1
   return counts


def map_read_pair(af, ar):
   rs = set()
   if af and ar and not af.aligned and not ar.aligned:
      return '_notaligned'
   if af and ar and not af.aQual < minaqual and ar.aQual < minaqual:
      return '_lowaqual'
   if af and af.aligned and af.aQual >= minaqual and af.iv.chrom in features.chrom_vectors.keys():
      for cigop in af.cigar:
         if cigop.type != "M":
            continue
         if reverse:
            cigop.ref_iv.strand = reverse_strand( cigop.ref_iv.strand )
         for iv, s in features[cigop.ref_iv].steps():
            rs = rs.union( s )
   if ar and ar.aligned and ar.aQual >= minaqual and ar.iv.chrom in features.chrom_vectors.keys():
      for cigop in ar.cigar:
         if cigop.type != "M":
            continue
         if not reverse:
            cigop.ref_iv.strand = reverse_strand( cigop.ref_iv.strand )
         for iv, s in features[cigop.ref_iv].steps():
               rs = rs.union( s )
   set_of_gene_names = set( [ f.name.split(":")[0] for f in rs ] )
   if len( set_of_gene_names ) == 0:
      return '_empty'
   elif len( set_of_gene_names ) > 1:
      return '_ambiguous'
   else:
      return rs


def clean_read_queue( queue, current_position ):
   clean_queue = dict( queue )
   for i in queue:
      if queue[i].mate_start.pos < current_position:
         warnings.warn( "Read "+ i + " claims to have an aligned mate that could not be found in the same chromosome." )
         del clean_queue[i]
   return clean_queue


if alignment == "sam":
   reader = HTSeq.SAM_Reader
else:
   if HTSeq.__version__ < '0.5.4p4':
      raise SystemError, "If you are using alignment files in a bam format, please update your HTSeq to 0.5.4p4 or higher"
   reader = HTSeq.BAM_Reader


# Now go through the aligned reads
num_reads = 0

if not is_PE:

   for a in reader( sam_file ):
      if a.optional_field("NH") > 1:
         continue
      if not a.aligned:
         counts[ '_notaligned' ] += 1
         continue
      if a.aQual < minaqual:
         counts[ '_lowaqual' ] += 1
         continue
      rs = set()
      for cigop in a.cigar:
         if cigop.type != "M":
            continue
         if reverse:
            cigop.ref_iv.strand = reverse_strand( cigop.ref_iv.strand )
         for iv, s in features[cigop.ref_iv].steps( ):
            rs = rs.union( s )
      set_of_gene_names = set( [ f.name.split(":")[0] for f in rs ] )
      if len( set_of_gene_names ) == 0:
         counts[ '_empty' ] += 1
      elif len( set_of_gene_names ) > 1:
         counts[ '_ambiguous' ] +=1
      else:
         for f in rs:
            counts[ f.name ] += 1
      num_reads += 1
      if num_reads % 100000 == 0:
         sys.stderr.write( "%d reads processed.\n" % num_reads )

else: # paired-end
   alignments = dict()
   if order == "name":
      for af, ar in HTSeq.pair_SAM_alignments( reader( sam_file ) ):
         if af == None or ar == None:
            continue
         elif ar.optional_field("NH") > 1 or af.optional_field("NH") > 1:
            continue
         elif af.iv.chrom != ar.iv.chrom:
            counts['_ambiguous_readpair_position'] += 1
            continue
         else:
            rs = map_read_pair( af, ar )
            counts = update_count_vector( counts, rs )
            num_reads += 1
         if num_reads % 100000 == 0:
            sys.stderr.write( "%d reads processed.\n" % num_reads )

   else:
      processed_chromosomes = dict()
      num_reads = 0
      current_chromosome=''
      current_position=''
      for a in reader( sam_file ):
         if not a.aligned:
            continue
         if a.optional_field("NH") > 1:
            continue
         if current_chromosome != a.iv.chrom:
            if current_chromosome in processed_chromosomes:
               raise SystemError, "A chromosome that had finished to be processed before was found again in the alignment file, is your alignment file properly sorted by position?"
            processed_chromosomes[current_chromosome] = 1
            alignments = clean_read_queue( alignments, current_position )
            del alignments
            alignments = dict()
         if current_chromosome == a.iv.chrom and a.iv.start < current_position:
            raise SystemError, "Current read position is smaller than previous reads, is your alignment file properly sorted by position?"
         current_chromosome = a.iv.chrom
         current_position = a.iv.start
         if a.read.name and a.mate_aligned:
            if a.read.name in alignments:
               b = alignments[ a.read.name ]
               if a.pe_which == "first" and b.pe_which == "second":
                  af=a
                  ar=b
               else:
                  af=b
                  ar=a
               rs = map_read_pair(af, ar)
               del alignments[ a.read.name ]
               counts = update_count_vector(counts, rs)
            else:
               if a.mate_start.chrom != a.iv.chrom:
                  counts['_ambiguous_readpair_position'] += 1
                  continue
               else:
                  alignments[ a.read.name ] = a
         else:
            continue
         num_reads += 1
         if num_reads % 200000 == 0:
            alignments = clean_read_queue( alignments, current_position )
            sys.stderr.write( "%d reads processed.\n" % (num_reads / 2) )



# Step 3: Write out the results

fout = open( out_file, "w" )
for fn in sorted( counts.keys() ):
   fout.write( "%s\t%d\n" % ( fn, counts[fn] ) )
fout.close()