Update db to 2023-11-09 (#16)

example_data/example_patient.capture.sbs_contributions.txt

@@ -0,0 +1,80 @@
+signature	contribution
+SBS1	0.003389830508474576
+SBS2	0.0
+SBS3	0.0
+SBS4	0.0
+SBS5	0.22372881355932203
+SBS6	0.0
+SBS7a	0.423728813559322
+SBS7b	0.0
+SBS7c	0.0
+SBS7d	0.0
+SBS8	0.0
+SBS9	0.0
+SBS10a	0.0
+SBS10b	0.11864406779661017
+SBS10c	0.0
+SBS10d	0.0
+SBS11	0.0
+SBS12	0.0
+SBS13	0.0
+SBS14	0.0
+SBS15	0.0
+SBS16	0.0
+SBS17a	0.0
+SBS17b	0.0
+SBS18	0.0
+SBS19	0.0
+SBS20	0.0
+SBS21	0.0
+SBS22	0.0
+SBS23	0.0
+SBS24	0.0
+SBS25	0.0
+SBS26	0.0
+SBS27	0.0
+SBS28	0.0
+SBS29	0.0
+SBS30	0.2305084745762712
+SBS31	0.0
+SBS32	0.0
+SBS33	0.0
+SBS34	0.0
+SBS35	0.0
+SBS36	0.0
+SBS37	0.0
+SBS38	0.0
+SBS39	0.0
+SBS40	0.0
+SBS41	0.0
+SBS42	0.0
+SBS43	0.0
+SBS44	0.0
+SBS45	0.0
+SBS46	0.0
+SBS47	0.0
+SBS48	0.0
+SBS49	0.0
+SBS50	0.0
+SBS51	0.0
+SBS52	0.0
+SBS53	0.0
+SBS54	0.0
+SBS55	0.0
+SBS56	0.0
+SBS57	0.0
+SBS58	0.0
+SBS59	0.0
+SBS60	0.0
+SBS84	0.0
+SBS85	0.0
+SBS86	0.0
+SBS87	0.0
+SBS88	0.0
+SBS89	0.0
+SBS90	0.0
+SBS91	0.0
+SBS92	0.0
+SBS93	0.0
+SBS94	0.0
+SBS95	0.0

example_data/example_patient.capture.somatic.indels.maf

@@ -1,3 +1,3 @@
-Hugo_Symbol	NCBI_Build	Chromosome	Start_position	End_position	Variant_Classification	Reference_Allele	Tumor_Seq_Allele1	Tumor_Seq_Allele2	Tumor_Sample_Barcode	Matched_Norm_Sample_Barcode	Annotation_Transcript	Protein_Change	t_alt_count	t_ref_count
-PMPCA	37	9	139312448	139312449	Intron	-	-	G	example_tumor_profile	example_normal_profile	ENST00000371717.3		31	92
-C10orf2	37	10	102748300	102748301	Frame_Shift_Del	TC	TC	-	example_tumor_profile	example_normal_profile	ENST00000370228.1	p.L112fs	28	294
+Hugo_Symbol	NCBI_Build	Chromosome	Start_position	End_position	Variant_Classification	Variant_Type	Reference_Allele	Tumor_Seq_Allele1	Tumor_Seq_Allele2	Tumor_Sample_Barcode	Matched_Norm_Sample_Barcode	Annotation_Transcript	Protein_Change	t_alt_count	t_ref_count
+PMPCA	37	9	139312448	139312449	Intron	INS	-	-	G	example_tumor_profile	example_normal_profile	ENST00000371717.3		31	92
+C10orf2	37	10	102748300	102748301	Frame_Shift_Del	DEL	TC	TC	-	example_tumor_profile	example_normal_profile	ENST00000370228.1	p.L112fs	28	294

moalmanac/annotator.py

@@ -177,14 +177,15 @@ class Almanac:
     implication = datasources.Almanac.implication
     implication_map = datasources.Almanac.implication_map
     description = datasources.Almanac.description
-    preferred_assertion = datasources.Almanac.preferred_assertion
 
     source_type = datasources.Almanac.source_type
     citation = datasources.Almanac.citation
     url = datasources.Almanac.url
     doi = datasources.Almanac.doi
     pmid = datasources.Almanac.pmid
     nct = datasources.Almanac.nct
+    publication_date = datasources.Almanac.publication_date
+    last_updated = datasources.Almanac.last_updated
 
     predictive_implication_map = datasources.Almanac.predictive_implication_map
 
@@ -820,8 +821,8 @@ def select_better_fusion_match(cls, list_of_matches):
 
     @classmethod
     def sort_and_subset_matches(cls, table, results_same_ontology, results_diff_ontology):
-        sort_columns = [cls.implication_map, cls.preferred_assertion]
-        sort_ascending = [False, False]
+        sort_columns = [cls.implication_map, cls.publication_date, cls.last_updated]
+        sort_ascending = [False, False, False]
         if results_same_ontology.any():
             results_same_ontology_sorted = cls.sort_dataframe(table[results_same_ontology], sort_columns, sort_ascending)
             results_same_ontology_records = cls.convert_dataframe_to_records(results_same_ontology_sorted)

moalmanac/colnames.ini

@@ -165,14 +165,15 @@ resistance = therapy_resistance
 prognosis = favorable_prognosis
 implication = predictive_implication
 implication_map = predictive_implication_map
-preferred_assertion = preferred_assertion
 connections = connections
 source_type = source_type
 citation = citation
 url = url
 doi = doi
 pmid = pmid
 nct = nct
+publication_date = publication_date
+last_updated = last_updated
 exac_common = exac_common
 exac_af = exac_af
 exac_ac = exac_ac
@@ -459,7 +460,6 @@ implication = ${maps:implication}
 implication_map = ${maps:implication_map}
 connections = ${maps:connections}
 description = ${maps:description}
-preferred_assertion = ${maps:preferred_assertion}
 sensitive_oncotree_code = ${maps:sensitive_oncotree_code}
 sensitive_oncotree_term = ${maps:sensitive_oncotree_term}
 sensitive_context = ${maps:sensitive_context}
@@ -523,6 +523,8 @@ url = ${maps:url}
 doi = ${maps:doi}
 pmid = ${maps:pmid}
 nct = ${maps:nct}
+publication_date = ${maps:publication_date}
+last_updated = ${maps:last_updated}
 chr = ${maps:chr}
 start = ${maps:start}
 end = ${maps:end}

moalmanac/config.ini

@@ -10,8 +10,8 @@ plot_mutational_signatures = on
 plot_preclinical_efficacy = on
 
 [versions]
-interpreter = 0.4.6
-database = v.2023-02-02
+interpreter = 0.4.7
+database = v.2023-11-09
 
 [exac]
 exac_common_af_threshold = 0.001

moalmanac/datasources.py

@@ -37,14 +37,15 @@ class Datasources:
     implication_map = COLNAMES[datasources_section]['implication_map']
     connections = COLNAMES[datasources_section]['connections']
     description = COLNAMES[datasources_section]['description']
-    preferred_assertion = COLNAMES[datasources_section]['preferred_assertion']
 
     source_type = COLNAMES[datasources_section]['source_type']
     citation = COLNAMES[datasources_section]['citation']
     url = COLNAMES[datasources_section]['url']
     doi = COLNAMES[datasources_section]['doi']
     pmid = COLNAMES[datasources_section]['pmid']
     nct = COLNAMES[datasources_section]['nct']
+    publication_date = COLNAMES[datasources_section]['publication_date']
+    last_updated = COLNAMES[datasources_section]['last_updated']
 
     sensitivity_matches = COLNAMES[datasources_section]['sensitivity_matches']
     resistance_matches = COLNAMES[datasources_section]['resistance_matches']
@@ -122,14 +123,15 @@ class Almanac:
     implication = Datasources.implication
     implication_map = Datasources.implication_map
     description = Datasources.description
-    preferred_assertion = Datasources.preferred_assertion
 
     source_type = Datasources.source_type
     citation = Datasources.citation
     url = Datasources.url
     doi = Datasources.doi
     pmid = Datasources.pmid
     nct = Datasources.nct
+    publication_date = Datasources.publication_date
+    last_updated = Datasources.last_updated
 
     sensitivity_matches = Datasources.sensitivity_matches
     resistance_matches = Datasources.resistance_matches

moalmanac/datasources/moalmanac/create_almanac_db.py

@@ -26,11 +26,11 @@ def create_config():
     'disease', 'context', 'oncotree_term', 'oncotree_code',
     'therapy_name', 'therapy_strategy', 'therapy_type',
     'therapy_sensitivity', 'therapy_resistance', 'favorable_prognosis',
-    'predictive_implication', 'description', 'last_updated', 'preferred_assertion'
+    'predictive_implication', 'description', 'last_updated'
 ]
 
 SOURCE_FIELDS = [
-    'source_type', 'citation', 'url', 'doi', 'pmid', 'nct', 'last_updated'
+    'source_type', 'citation', 'url', 'doi', 'pmid', 'nct', 'publication_date', 'last_updated'
 ]
 
 VARIANT_FIELDS = [
@@ -59,7 +59,7 @@ def create_config():
 ]
 
 MUTATIONAL_SIGNATURE_FIELDS = [
-    'cosmic_signature_number', 'cosmic_signature_version'
+    'cosmic_signature'
 ]
 
 MICROSATELLITE_FIELDS = [
@@ -86,7 +86,7 @@ def copy_number(cls, record):
 
     @classmethod
     def cosmic_mutational_signature(cls, record):
-        return f"Cosmic signature (version {record['cosmic_signature_version']}) {record['cosmic_signature_number']}"
+        return f"Cosmic signature {record['cosmic_signature']} (version 3.4)"
 
     @classmethod
     def germline_variant(cls, record):