October gene table update: v7

gene-table-update/build-input-for-importer/Readme.md

@@ -10,7 +10,7 @@ Scripts to build an input file, to be used by importer to build/update seedDB ge
 #### Step 1 - Download latest HGNC gene table
 
 ```
-wget ftp://ftp.ebi.ac.uk/pub/databases/genenames/hgnc/tsv/hgnc_complete_set.txt
+wget https://storage.googleapis.com/public-download-files/hgnc/archive/archive/monthly/tsv/hgnc_complete_set_2025-10-07.txt
 ```
 
 OR

gene-table-update/build-input-for-importer/build-gene-table-input.py

@@ -39,12 +39,12 @@ def format(self, record):
         return f'<div style="color:{message_color};"><strong style="color:{levelname_color};">[{levelname}]</strong>: {message}</div>'
 
 # Define paths for mappings and supplemental files
-TYPE_MAPPING_PATH = "/Users/bsatravada/Desktop/tets_hgnc_build/build-input-for-importer/mappings/type-mapping.txt"
-LOCATION_MAPPING_PATH = "/Users/bsatravada/Desktop/tets_hgnc_build/build-input-for-importer/mappings/location-mapping.txt"
-MAIN_SUPP_PATH = "/Users/bsatravada/Desktop/tets_hgnc_build/build-input-for-importer/supp-files/main-supp/complete-supp-main.txt"
-ALIAS_SUPP_PATH = "/Users/bsatravada/Desktop/tets_hgnc_build/build-input-for-importer/supp-files/alias-supp.txt"
-ENTREZ_ID_SUPP_PATH = "/Users/bsatravada/Desktop/tets_hgnc_build/build-input-for-importer/supp-files/entrez-id-supp.txt"
-LOCATION_SUPP_PATH = "/Users/bsatravada/Desktop/tets_hgnc_build/build-input-for-importer/supp-files/location-supp.txt"
+TYPE_MAPPING_PATH = "/Users/bsatravada/Desktop/HGNC_OCT_2025/build-input-for-importer/mappings/type-mapping.txt"
+LOCATION_MAPPING_PATH = "/Users/bsatravada/Desktop/HGNC_OCT_2025/build-input-for-importer/mappings/location-mapping.txt"
+MAIN_SUPP_PATH = "/Users/bsatravada/Desktop/HGNC_OCT_2025/build-input-for-importer/supp-files/main-supp/complete-supp-main.txt"
+ALIAS_SUPP_PATH = "/Users/bsatravada/Desktop/HGNC_OCT_2025/build-input-for-importer/supp-files/alias-supp.txt"
+ENTREZ_ID_SUPP_PATH = "/Users/bsatravada/Desktop/HGNC_OCT_2025/build-input-for-importer/supp-files/entrez-id-supp.txt"
+LOCATION_SUPP_PATH = "/Users/bsatravada/Desktop/HGNC_OCT_2025/build-input-for-importer/supp-files/location-supp.txt"
 
 gene_dict = {}
 
@@ -129,13 +129,13 @@ def cleanup_entrez_id(_input_file_name):
     return _exiting_flag
 
 # Removes all entries with locus_type value as RNA, micro
-def remove_mirna():
-    logging.info("Removing miRNA entries")
-    for _key in list(gene_dict.keys()):
-        _gene_obj = gene_dict[_key]
-        if _gene_obj["locus_type"] == "RNA, micro":
-            del gene_dict[_key]
-    logging.info("Finished removing miRNA entries")
+# def remove_mirna():
+#     logging.info("Removing miRNA entries")
+#     for _key in list(gene_dict.keys()):
+#         _gene_obj = gene_dict[_key]
+#         if _gene_obj["locus_type"] == "RNA, micro":
+#             del gene_dict[_key]
+#     logging.info("Finished removing miRNA entries")
 
 
 # Merge values locus_group and locus_type into one column type

gene-table-update/build-input-for-importer/supp-files/entrez-id-supp.txt

@@ -1,85 +1,90 @@
-hugo_gene_symbol	assigned_entrez_id
-AKAP2	R
-BACH1-IT1	R
-CYP2A7P2	1550
-DDX11L6	R
-DDX11L7	R
-DDX11L13	R
-DDX11L15	R
-DNAJB5P1	R
-EBVM1	R
-ERVH-8	R
-ERVH-9	R
-ERVH-10	R
-FRA3B	R
-GTF2IP16	R
-GTF2IP21	R
-MT-7SDNA	R
-MT-ATT	R
-MT-CSB1	R
-MT-CSB2	R
-MT-CSB3	R
-MT-HPR	R
-MT-HSP1	R
-MT-HSP2	R
-MT-LIPCAR	R
-MT-LSP	R
-MT-OHR	R
-MT-OLR	R
-MT-RNR3	R
-MT-TAS	R
-MT-TER	R
-MT-TFH	R
-MT-TFL	R
-MT-TFX	R
-MT-TFY	R
-MTND6P36	R
-PALM2	R
-SLC7A2-IT1	R
-TRA-AGC13-3	R
-TRE-TTC8-2	R
-TRMT10BP1	101060002
-TRN-GTT15-2	R
-TRN-GTT19-2	R
-TRQ-CTG8-3	R
-TRR-CCT6-1	R
-MTND4P38	107133487
-C11orf98P3	127138860
-CCDC86-AS1	127138859
-LINC03051	127138867
-LINC03052	127138865
-SMIM46	127138866
-SYT2-AS1	129810494
-NUPR2P1	129810496
-LINC03100	129810495
-MYADM-AS2	129810493
-RNU6-840P	131183827
-ENEMAL	129810501
-PCHILR	129810500
-ALG14-AS1	137805484
-GAD1-AS1	137805480
-GLYATL1-AS1	137805481
-LINC03130	137805482
-LINC03137	137805492
-LINC03138	137805490
-NQO1-AS1	137933868
-PEARL1	138349443
-PROK2-DT	137805491
-PTPN11BP	137805489
-RBBP8-AS1	137805483
-SCDAL	137805493
-SYT1-AS1	137805486
-TGILR	138349442
-CRK-AS1	137805487
-CYP3A4-AS1	137805488
-DCAF12-AS1	137805485
-ADAMTS16-AS1	137805484
-ASAR1-1	R
-ASAR6-1	R
-ASAR6-2	R
-ASAR8-1	R
-ASAR9-1	R
-ASAR9-2	R
-ASAR9-3	R
-ASAR15-1	R
-CHILL1	127460870
+hugo_gene_symbol	assigned_entrez_id
+AKAP2	R
+BACH1-IT1	R
+CYP2A7P2	1550
+DDX11L6	R
+DDX11L7	R
+DDX11L13	R
+DDX11L15	R
+DNAJB5P1	R
+EBVM1	R
+ERVH-8	R
+ERVH-9	R
+ERVH-10	R
+FRA3B	R
+GTF2IP16	R
+GTF2IP21	R
+MT-7SDNA	R
+MT-ATT	R
+MT-CSB1	R
+MT-CSB2	R
+MT-CSB3	R
+MT-HPR	R
+MT-HSP1	R
+MT-HSP2	R
+MT-LIPCAR	R
+MT-LSP	R
+MT-OHR	R
+MT-OLR	R
+MT-RNR3	R
+MT-TAS	R
+MT-TER	R
+MT-TFH	R
+MT-TFL	R
+MT-TFX	R
+MT-TFY	R
+MTND6P36	R
+PALM2	R
+SLC7A2-IT1	R
+TRA-AGC13-3	R
+TRE-TTC8-2	R
+TRMT10BP1	101060002
+TRN-GTT15-2	R
+TRN-GTT19-2	R
+TRQ-CTG8-3	R
+TRR-CCT6-1	R
+MTND4P38	107133487
+C11orf98P3	127138860
+CCDC86-AS1	127138859
+LINC03051	127138867
+LINC03052	127138865
+SMIM46	127138866
+SYT2-AS1	129810494
+NUPR2P1	129810496
+LINC03100	129810495
+MYADM-AS2	129810493
+RNU6-840P	131183827
+ENEMAL	129810501
+PCHILR	129810500
+ALG14-AS1	137805484
+GAD1-AS1	137805480
+GLYATL1-AS1	137805481
+LINC03130	137805482
+LINC03137	137805492
+LINC03138	137805490
+NQO1-AS1	137933868
+PEARL1	138349443
+PROK2-DT	137805491
+PTPN11BP	137805489
+RBBP8-AS1	137805483
+SCDAL	137805493
+SYT1-AS1	137805486
+TGILR	138349442
+CRK-AS1	137805487
+CYP3A4-AS1	137805488
+DCAF12-AS1	137805485
+ADAMTS16-AS1	137805484
+ASAR1-1	R
+ASAR6-1	R
+ASAR6-2	R
+ASAR8-1	R
+ASAR9-1	R
+ASAR9-2	R
+ASAR9-3	R
+ASAR15-1	R
+CHILL1	127460870
+ABLIM2-DT	R
+LINC03160	R
+PMAR72	R
+STEAP1B-AS2	R
+S45-AS1	R