Feature/pep evidences df

src/pyOpenMS/pyopenms/dataframes.py

@@ -293,20 +293,26 @@ def get_key(val):
     cv.loadFromOBO("psims", File.getOpenMSDataPath() + "/CV/psi-ms.obo")
     clearMVs = [cv.getTerm(m).name if m.startswith("MS:") else m for m in decodedMVs]
     #cols = ["id", "RT", "mz", "score", "charge"] + decodedMVs
-    clearcols = ["id", "RT", "mz", mainscorename, "charge"] + clearMVs
-    coltypes = ['U100', 'f', 'f', 'f', 'i'] + types
+    clearcols = ["id", "RT", "mz", mainscorename, "charge", "protein_accession", "start", "end"] + clearMVs
+    coltypes = ['U100', 'f', 'f', 'f', 'i','U100', 'U100', 'U100'] + types
     dt = list(zip(clearcols, coltypes))
 
     def extract(pep):
         hits = pep.getHits()
         if not hits:
             if export_unidentified:
-                return tuple(pep.getIdentifier().encode('utf-8'), pep.getRT(), pep.getMZ(), default_missing_values[float], default_missing_values[int], *dmv)
+                return (pep.getIdentifier().encode('utf-8'), pep.getRT(), pep.getMZ(), default_missing_values[float], default_missing_values[int],
+                        default_missing_values[str], default_missing_values[str], default_missing_values[str], *dmv)
             else:
                 return
 
         besthit = hits[0]
-        ret = [pep.getIdentifier().encode('utf-8'), pep.getRT(), pep.getMZ(), besthit.getScore(), besthit.getCharge()]
+        ret = [pep.getIdentifier().encode('utf-8'), pep.getRT(), pep.getMZ(), besthit.getScore(), besthit.getCharge()] 
+        # add accession, start and end positions of peptide evidences as comma separated str (like in mzTab)
+        evs = besthit.getPeptideEvidences()
+        ret += [','.join(v) if v else default_missing_values[str] for v in ([e.getProteinAccession() for e in evs],
+                                                                            [str(e.getStart()) for e in evs],
+                                                                            [str(e.getEnd()) for e in evs])]
         for k in metavals:
             if besthit.metaValueExists(k):
                 val = besthit.getMetaValue(k)

src/pyOpenMS/tests/unittests/test000.py

@@ -2255,12 +2255,31 @@ def test_peptide_identifications_to_df():
     h.setCharge(2)
     h.setMetaValue("StringMetaValue", "Value")
     h.setMetaValue("IntMetaValue", 2)
+    e1 = pyopenms.PeptideEvidence()
+    e1.setProteinAccession("sp|Accession1")
+    e1.setStart(123)
+    e1.setEnd(141)
+    e2 = pyopenms.PeptideEvidence()
+    e2.setProteinAccession("sp|Accession2")
+    e2.setStart(12)
+    e2.setEnd(24)
+    h.setPeptideEvidences([e1, e2])
     p.insertHit(h)
 
     peps.append(p)
-    peps.append(p)
 
-    assert pyopenms.peptide_identifications_to_df(peps).shape == (2,7)
+    p1 = pyopenms.PeptideIdentification()
+    p1.setRT(1243.56)
+    p1.setMZ(240.0)
+    p1.setScoreType("ScoreType")
+    p1.setHigherScoreBetter(False)
+    p1.setIdentifier("IdentificationRun2")
+
+    peps.append(p1)
+
+    assert pyopenms.peptide_identifications_to_df(peps).shape == (2,10)
+    assert pyopenms.peptide_identifications_to_df(peps)['protein_accession'][0] == 'sp|Accession1,sp|Accession2'
+    assert pyopenms.peptide_identifications_to_df(peps, export_unidentified=False).shape == (1,10)
 
 @report
 def testPepXMLFile():