Merge pull request #202 from Michael-C-Strobel/Massbank-Update

mwang87 · web-flow · commit 302d7f75b966 · 2024-12-02T11:40:15.000-08:00
Allow lower case letters in Massbank identifiers.
diff --git a/metabolomics_spectrum_resolver/parsing.py b/metabolomics_spectrum_resolver/parsing.py
@@ -416,7 +416,7 @@ def _parse_massbank(usi: str) -> Tuple[sus.MsmsSpectrum, str]:
     index = match.group(4)
     # Clean up the new MassBank accessions if necessary.
     massbank_accession = re.match(
-        r"MSBNK-[A-Z0-9_]{1,32}-([A-Z0-9_]{1,64})", index
+        r"MSBNK-[A-z0-9_]{1,32}-([A-Za-z0-9_]{1,64})", index
     )
     if massbank_accession is not None:
         index = massbank_accession.group(1)
diff --git a/test/usi_test_data.py b/test/usi_test_data.py
@@ -10,6 +10,8 @@
     "mzspec:GNPS:GNPS-LIBRARY:accession:CCMSLIB00005436077",
     "mzspec:MASSBANK::accession:SM858102",
     "mzspec:MASSBANK::accession:MSBNK-AAFC-AC000646",
+    # New Massbank identifier with lowercase
+    "mzspec:MASSBANK::accession:MSBNK-Athens_Univ-AU259904",
     "mzspec:MS2LDA:TASK-190:accession:270684",
     "mzspec:MOTIFDB::accession:171163",
     "mzspec:MSV000082791:(-)-epigallocatechin:scan:2",

Original file line number	Diff line number	Diff line change
`@@ -416,7 +416,7 @@ def _parse_massbank(usi: str) -> Tuple[sus.MsmsSpectrum, str]:`
`416`	`416`	`index = match.group(4)`
`417`	`417`	`# Clean up the new MassBank accessions if necessary.`
`418`	`418`	`massbank_accession = re.match(`
`419`		`- r"MSBNK-[A-Z0-9_]{1,32}-([A-Z0-9_]{1,64})", index`
	`419`	`+ r"MSBNK-[A-z0-9_]{1,32}-([A-Za-z0-9_]{1,64})", index`
`420`	`420`	`)`
`421`	`421`	`if massbank_accession is not None:`
`422`	`422`	`index = massbank_accession.group(1)`