Merge pull request #113 from singjc/main

Add: Ion Mobility data to Sage Reader

Author: RalfG

psm_utils/io/sage.py

@@ -57,6 +57,14 @@ def _get_peptide_spectrum_match(self, psm_dict) -> PSM:
                 rescoring_features[ft] = psm_dict[ft]
             except KeyError:
                 continue
+        
+        # If ion mobility is not 0.0 (not present), add it to the rescoring features     
+        if float(psm_dict['ion_mobility']):
+            rescoring_features.update({
+            'ion_mobility': float(psm_dict['ion_mobility']),
+            'predicted_mobility': float(psm_dict['predicted_mobility']),
+            'delta_mobility': float(psm_dict['delta_mobility'])
+            })
 
         return PSM(
             peptidoform=self._parse_peptidoform(
@@ -70,6 +78,7 @@ def _get_peptide_spectrum_match(self, psm_dict) -> PSM:
             score=float(psm_dict[self.score_column]),
             precursor_mz=self._parse_precursor_mz(psm_dict["expmass"], psm_dict["charge"]),
             retention_time=float(psm_dict["rt"]),
+            ion_mobility=float(psm_dict["ion_mobility"]) if float(psm_dict["ion_mobility"]) else None,
             protein_list=psm_dict["proteins"].split(";"),
             source="sage",
             rank=int(float(psm_dict["rank"])),

tests/test_data/resultsIM.sage.tsv

@@ -0,0 +1,2 @@
+psm_id	peptide	proteins	num_proteins	filename	scannr	rank	label	expmass	calcmass	charge	peptide_len	missed_cleavages	semi_enzymatic	isotope_error	precursor_ppm	fragment_ppm	hyperscore	delta_next	delta_best	rt	aligned_rt	predicted_rt	delta_rt_model	ion_mobility	predicted_mobility	delta_mobility	matched_peaks	longest_b	longest_y	longest_y_pct	matched_intensity_pct	scored_candidates	poisson	sage_discriminant_score	posterior_error	spectrum_q	peptide_q	protein_q	ms2_intensity
+529791	YVDDTQFVRFDSDAASPR	sp|P01889|HLAB_HUMAN;sp|P10321|HLAC_HUMAN	2	G220824_028_Slot2-34_1_6753.mzml	index=45761	1	1	2086.9507	2087.9548	3	18	0	0	-1.00335	0.38332784	1.5193833	44.947527657385436	24.626872218188044	0.0	23.004148	0.5305706	0.5301316	0.0004390478	0.96470714	0.9236175	0.041089654	17	2	6	0.33333334	46.215378	33149	-14.997740845471464	0.47067946	-96.47973	0.0000107030855	0.00009881538	0.0006954081	26230.0

tests/test_data/test_out_sage.idXML

@@ -27,6 +27,7 @@
 				<UserParam type="float" name="delta_rt_model" value="0.993444"/>
 				<UserParam type="float" name="aligned_rt" value="0.993444"/>
 				<UserParam type="float" name="predicted_rt" value="0.0"/>
+				<UserParam type="float" name="predicted_mobility" value="0.0"/>
 				<UserParam type="float" name="matched_peaks" value="22.0"/>
 				<UserParam type="float" name="longest_b" value="9.0"/>
 				<UserParam type="float" name="longest_y" value="12.0"/>

tests/test_io/test_idxml.py

@@ -107,7 +107,7 @@ def test_write_file_with_pyopenms_objects(self):
         assert sha == expected_sha
 
     def test_write_file_without_pyopenms_objects(self):
-        expected_sha = "148889926276fbe391e23ed7952c3a8410fc67ffb099bbf1a72df75f8d727ccd"
+        expected_sha = "148889926276fbe391e23ed7952c3a8410fc67ffb099bbf1a72df75f8d727ccd" #TODO: can cause problems locally depending on dependency versions
         reader = SageTSVReader("./tests/test_data/results.sage.tsv")
         psm_list = reader.read_file()
         writer = IdXMLWriter("./tests/test_data/test_out_sage.idXML")

tests/test_io/test_sage.py

@@ -47,13 +47,62 @@
     },
 )
 
+test_psm_im = PSM(
+    peptidoform="YVDDTQFVRFDSDAASPR/3",
+    spectrum_id="index=45761",
+    run="G220824_028_Slot2-34_1_6753",
+    collection=None,
+    spectrum=None,
+    is_decoy=False,
+    score=0.47067946,
+    qvalue=0.0000107030855,
+    pep=None,
+    precursor_mz=696.6580583654032,
+    retention_time=23.004148,
+    ion_mobility=0.96470714,
+    protein_list=['sp|P01889|HLAB_HUMAN','sp|P10321|HLAC_HUMAN'],
+    rank=1,
+    source="sage",
+    metadata={},
+    rescoring_features={
+        "expmass": 2086.9507,
+        "calcmass": 2087.9548,
+        "peptide_len": 18.0,
+        "missed_cleavages": 0.0,
+        "isotope_error": -1.00335,
+        "precursor_ppm": 0.38332784,
+        "fragment_ppm": 1.5193833,
+        "hyperscore": 44.947527657385436,
+        "delta_next": 24.626872218188044,
+        "delta_best": 0.0,
+        "delta_rt_model": 0.0004390478,
+        "aligned_rt": 0.5305706,
+        "predicted_rt": 0.5301316,
+        "matched_peaks": 17.0,
+        "longest_b": 2.0,
+        "longest_y": 6.0,
+        "longest_y_pct": 0.33333334,
+        "matched_intensity_pct": 46.215378,
+        "scored_candidates": 33149.0,
+        "poisson": -14.997740845471464,
+        "ms2_intensity": 26230.0,       
+        "ion_mobility": 0.96470714,
+        "predicted_mobility": 0.9236175,
+        "delta_mobility": 0.041089654,
+    }
+)
+
 
 class TestSageTSVReader:
     def test_iter(self):
         with SageTSVReader("./tests/test_data/results.sage.tsv") as reader:
             for psm in reader:
                 psm.provenance_data = {}
                 assert psm == test_psm
+        with SageTSVReader("./tests/test_data/resultsIM.sage.tsv") as reader:
+            for psm in reader:
+                psm.provenance_data = {}
+                assert psm == test_psm_im
 
 
 class TestSageParquetReader: