Changeset 45

Timestamp:

Aug 11, 2010, 4:10:19 PM (10 years ago)

Author:

j@…

bzr:base-revision:

j@dannynavarro.net-20100809141712-f1elceltr5n95752

bzr:committer:

Danny Navarro <j@dannynavarro.net>

bzr:file-ids:

mzcms/models.py models.py-20100730084238-fjjwldiefr0w07zv-4
mzcms/parsers.py parsers.py-20100806092910-g1sxvv1o5b9umkof-1

bzr:mapping-version:

v4

bzr:repository-uuid:

724254b2-fbe6-419d-9466-c04ef4c9d29d

bzr:revision-id:

j@dannynavarro.net-20100809152143-wupx7dm9da51912j

bzr:revno:

45

bzr:revprop:branch-nick:

trunk

bzr:root:

trunk

bzr:timestamp:

2010-08-09 17:21:43.714999914 +0200

bzr:user-agent:

bzr2.1.2+bzr-svn1.0.3

svn:original-date:

2010-08-09T15:21:43.715000Z

Message:

Parsing spectra sucessfully

Location:

trunk/mzcms

Files:

• models.py (modified) (1 diff)
• parsers.py (modified) (5 diffs)

trunk/mzcms/models.py

@@ -48 +48 @@
     def __init__(self, scan, run, peaks, prec_mz, prec_charge):
         self.scan = int(scan)
-        self.run = str(scan)
+        self.run = str(run)
         self.peaks = tuple(peaks)
         self.prec_mz = float(prec_mz)

trunk/mzcms/parsers.py

@@ -11 +11 @@
 from collections import defaultdict
 from collections import namedtuple
+from itertools import izip
 
 SEPARATOR = '--gc0p4Jq0M2Yt08jU534c0p'
@@ -46 +47 @@
     """
     def __init__(self,
+                 protein_factory,
+                 peptide_factory,
+                 spectrum_factory,
+                 psm_factory,
                  scan_str=r'FinneganScanNumber%3a%20(\d+)%20',
                  rawfn_str=r'RawFile%3a%20(.+raw)',
@@ -51 +56 @@
                  contaminant_str=r'^IPI:CON_:IPI',
                  # XXX: Fix default factories
-                 protein_factory=dict,
-                 peptide_factory=dict,
-                 spectrum_factory=dict,
-                 psm_factory=dict
                  ):
         self.scan_regex = re.compile(scan_str)
@@ -62 +63 @@
         self.non_target_regexes = (contaminant_regex, decoy_regex)
         self.rawfn_regex = re.compile(rawfn_str)
+        self.protein_factory = protein_factory
+        self.peptide_factory = protein_factory
+        self.spectrum_factory=spectrum_factory
+        self.psm_factory=psm_factory
 
     def _parse_spectra_fn(self, dat_file):
@@ -168 +173 @@
         nativeid_psms = self._parse_psms(dat_file)
         extraspec = self._parse_extraspec(dat_file)
-        #for native_id, psms in nativeid_psms.items():
-        #    for psm in psms:
-        #        psm['#SpectraFile'] = pkl_fn
-        #        psm['FragMode'] = frag_mode
-        #        psm.update(nativeid_spectra[native_id])
-        #        yield psm
-        #return proteins, peptides, spectra, psms
+        spectra = dict()
+        for summ, extra in izip(summspec, extraspec):
+            spectrum = self.spectrum_factory(prec_mz=summ[0],
+                                             prec_charge=summ[1],
+                                             run=extra[0],
+                                             scan=extra[1],
+                                             peaks="TBI")
+            spectra[(spectrum.scan, spectrum.run)] = spectrum
+        import ipdb; ipdb.set_trace()
+        return proteins, peptides, spectra, psms
 
 def get_prot_ids(accs):