#42 - Ensembl T2T

generate_transcript_data/all_transcripts.sh

@@ -34,6 +34,11 @@ cd GRCh38
 ${BASE_DIR}/refseq_transcripts_grch38.sh
 cd ..
 
+mkdir -p CHM13v2.0
+cd CHM13v2.0
+${BASE_DIR}/refseq_transcripts_chm13v2.sh
+cd ..
+
 # Combine genome builds (we're in refseq dir)
 REFSEQ_COMBO=cdot-${CDOT_VERSION}.refseq.grch37_grch38.json.gz
 if [[ ! -e ${REFSEQ_COMBO} ]]; then
@@ -43,11 +48,6 @@ if [[ ! -e ${REFSEQ_COMBO} ]]; then
       --output ${REFSEQ_COMBO}
 fi
 
-mkdir -p CHM13v2.0
-cd CHM13v2.0
-${BASE_DIR}/refseq_transcripts_chm13v2.sh
-cd ..
-
 cd ..
 
 # Ensembl
@@ -64,6 +64,12 @@ cd GRCh38
 ${BASE_DIR}/ensembl_transcripts_grch38.sh
 cd ..
 
+mkdir -p CHM13v2.0
+cd CHM13v2.0
+${BASE_DIR}/ensembl_transcripts_chm13v2.sh
+cd ..
+
+
 # Combine genome builds (we're in ensembl dir)
 ENSEMBL_COMBO=cdot-${CDOT_VERSION}.ensembl.grch37_grch38.json.gz
 if [[ ! -e ${ENSEMBL_COMBO} ]]; then

generate_transcript_data/ensembl_transcripts_chm13v2.sh

@@ -0,0 +1,31 @@
+#!/bin/bash
+
+set -e
+
+BASE_DIR=$(dirname ${BASH_SOURCE[0]})
+CDOT_VERSION=$(${BASE_DIR}/cdot_json.py --version)
+
+if [[ -z ${GENE_INFO_JSON} ]]; then
+  echo "You need to set environment variable GENE_INFO_JSON, pointing to the filename produced by cdot_gene_info.py"
+  exit 1
+fi
+
+merge_args=()
+for release in 2022_06 2022_07; do
+  filename=Homo_sapiens-GCA_009914755.4-${release}-genes.gff3.gz
+  url=https://ftp.ensembl.org/pub/rapid-release/species/Homo_sapiens/GCA_009914755.4/ensembl/geneset/${release}/${filename}
+  cdot_file=cdot-${CDOT_VERSION}.$(basename $filename .gz).json.gz
+
+  if [[ ! -e ${filename} ]]; then
+    wget ${url}
+  fi
+  if [[ ! -e ${cdot_file} ]]; then
+    ${BASE_DIR}/cdot_json.py gff3_to_json "${filename}" --url "${url}" --genome-build=CHM13v2.0 --output "${cdot_file}" --gene-info-json="${GENE_INFO_JSON}"
+  fi
+  merge_args+=(${cdot_file})
+done
+
+merged_file="cdot-${CDOT_VERSION}.ensembl.CHM13v2.0.json.gz"
+if [[ ! -e ${merged_file} ]]; then
+  ${BASE_DIR}/cdot_json.py merge_historical ${merge_args[@]} --genome-build=CHM13v2.0 --output "${merged_file}"
+fi