temp version.. not working yet.. just saving

balajipandian · Jan 22, 2016 · 85257ce · 85257ce
1 parent 875ab11
commit 85257ce
Show file tree

Hide file tree

Showing 19 changed files with 7,076 additions and 203 deletions.
diff --git a/setup.py b/setup.py
@@ -19,8 +19,11 @@
 
 
 extensions = [
-    Extension('taco.lib.cbedgraph',
-              sources=['taco/lib/cbedgraph.pyx'],
+    Extension('taco.lib.cBedGraph',
+              sources=['taco/lib/cBedGraph.pyx'],
+              include_dirs=[numpy_inc]),
+    Extension('taco.lib.cChangePoint',
+              sources=['taco/lib/cChangePoint.pyx'],
               include_dirs=[numpy_inc])
 ]
 
@@ -31,7 +34,7 @@ def main():
           description='transcriptome meta-assembly from rna-seq',
           author='Matthew Iyer and Yashar Niknafs',
           author_email='[email protected]',
-          requires=['numpy', 'networkx'],
+          requires=['numpy', 'networkx', 'h5py'],
           license='GPL',
           platforms='Linux',
           url='https://github.com/yniknafs/taco',

diff --git a/taco/lib/assemble.py b/taco/lib/assemble.py
@@ -67,7 +67,11 @@ def get_splice_graphs(locus):
     # return transcript_graphs
 
 
-def assemble(gtf_file, output_dir,
+def assemble(gtf_file,
+             expr_h5_file,
+             chrom_sizes_file,
+             node_gtf_file,
+             output_dir,
              guided_strand=False,
              guided_ends=False,
              guided_assembly=False):
@@ -76,6 +80,10 @@ def assemble(gtf_file, output_dir,
     raw_bgfilehd = Locus.open_bedgraphs(file_prefix)
     file_prefix = os.path.join(output_dir, 'loci.resolved')
     resolved_bgfilehd = Locus.open_bedgraphs(file_prefix)
+    # setup expression hdf5
+    expr_h5f = Locus.open_expression_hdf5(expr_h5_file, chrom_sizes_file)
+    # setup node gtf file
+    node_fileh = open(node_gtf_file, 'w')
 
     # parse gtf file
     for interval, gtf_lines in GTF.parse_loci(open(gtf_file)):
@@ -86,9 +94,9 @@ def assemble(gtf_file, output_dir,
         logging.debug('Locus %s:%d-%d: '
                       '%d transfrags (+: %d, -: %d, .: %d)' %
                       (chrom, start, end, len(t_dict),
-                       len(locus.strand_transfrags[Strand.POS]),
-                       len(locus.strand_transfrags[Strand.NEG]),
-                       len(locus.strand_transfrags[Strand.NA])))
+                       len(locus.get_transfrags(Strand.POS)),
+                       len(locus.get_transfrags(Strand.NEG)),
+                       len(locus.get_transfrags(Strand.NA))))
         # write raw bedgraph files
         locus.write_bedgraph(raw_bgfilehd)
         # resolve unstranded transcripts
@@ -97,16 +105,27 @@ def assemble(gtf_file, output_dir,
             logging.debug('Locus %s:%d-%d: %d '
                           'resolved (+: %d, -: %d, .: %d)' %
                           (chrom, start, end, num_resolved,
-                           len(locus.strand_transfrags[Strand.POS]),
-                           len(locus.strand_transfrags[Strand.NEG]),
-                           len(locus.strand_transfrags[Strand.NA])))
+                           len(locus.get_transfrags(Strand.POS)),
+                           len(locus.get_transfrags(Strand.NEG)),
+                           len(locus.get_transfrags(Strand.NA))))
         # write bedgraph files after strand resolved
         locus.write_bedgraph(resolved_bgfilehd)
+        locus.write_expression_hdf5(expr_h5f)
 
         # create splice graphs
-        for sg in get_splice_graphs(locus):
-            print 'splice graph', sg.chrom, Strand.to_gtf(sg.strand), sg.start, sg.end
-            pass
+        for slocus in locus.L:
+            if slocus is None:
+                continue
+            G = slocus.create_splice_graph()
+            print len(G)
+
+        # for sg in get_splice_graphs(locus):
+        #    pass
+#            for line in sg.get_change_point_data():
+#                print >>changept_fileh, line
+
+    # close expression
+    expr_h5f.close()
 
     # close bedgraph files
     Locus.close_bedgraphs(raw_bgfilehd)

diff --git a/taco/lib/base.py b/taco/lib/base.py
@@ -8,7 +8,6 @@
 import collections
 import logging
 import os
-import numpy as np
 
 __author__ = "Matthew Iyer and Yashar Niknafs"
 __copyright__ = "Copyright 2015"
@@ -23,15 +22,15 @@
 class TacoError(Exception):
     pass
 
-FLOAT_DTYPE = np.float64
-
 Exon = collections.namedtuple('Exon', ['start', 'end'])
 
 
 class Strand:
     POS = 0
     NEG = 1
     NA = 2
+
+    STRANDS = [POS, NEG, NA]
     NAMES = ['pos', 'neg', 'none']
     FROM_GTF = {'+': POS, '-': NEG, '.': NA}
     TO_GTF = {POS: '+', NEG: '-', NA: '.'}

diff --git a/taco/lib/bedgraph.py b/taco/lib/bedgraph.py
@@ -4,7 +4,7 @@
 import collections
 import numpy as np
 
-from taco.lib.base import FLOAT_DTYPE
+from taco.lib.dtypes import FLOAT_DTYPE
 
 __author__ = "Matthew Iyer and Yashar Niknafs"
 __copyright__ = "Copyright 2016"