feat: add appris data (#132)

gencode v25, 26, 27, vM15

also: fix compile error
see https://github.com/deepgenomics/GenomeKit/actions/runs/13533190669/job/37819749880?pr=132

conda-recipe/conda_build_config.yaml

@@ -4,7 +4,7 @@ c_stdlib:
 - sysroot # [linux]
 - macosx_deployment_target # [osx]
 c_stdlib_version:
-- '2.12' # [linux]
+- '2.17' # [linux]
 - '11.0' # [osx]
 python:
 - '3.9'

genome_kit/_build_appris.py

@@ -52,12 +52,12 @@
 # hg38
 REMOTE_FILES.update(("gencode.v{version}{lift}{basic}".format(version=version, lift=lift, basic=basic),
                      PATH_FMT.format(species="homo_sapiens", data_version=data_version))
-                    for version, data_version in [(29, "e94v28"), (41, "e103v45")]
+                    for version, data_version in [(25, "e87v22"), (26, "e88v22"), (27, "e91v27"), (29, "e94v28"), (41, "e103v45")]
                     for lift in ["", "lift37"] for basic in ["", ".basic"])
 # mouse
 REMOTE_FILES.update(("gencode.vM{version}{basic}".format(version=version, basic=basic),
                      PATH_FMT.format(species="mus_musculus", data_version=data_version))
-                    for version, data_version in [(19, "e94v28"), (30, "e107v47")] for basic in ["", ".basic"])
+                    for version, data_version in [(15, "e88v22"), (19, "e94v28"), (30, "e107v47")] for basic in ["", ".basic"])
 # ncbi refseq
 REMOTE_FILES.update(("ncbi_refseq.{version}".format(version=version),
                      PATH_FMT.format(species="homo_sapiens", data_version=data_version))

genome_kit/_gk_data_config.py

@@ -41,6 +41,8 @@ def get_appris_version(annotation):
         return "2022_07.v47"
     elif any(x in annotation for x in ["ncbi_refseq.v110"]):
         return "2023_08.v48"
+    elif any(x in annotation for x in ["gencode.v27"]):
+        return "2018_02.v27"
     return "2017_06.v23"