From ed8613310d023699a1ea39694fc645e4634d1a8c Mon Sep 17 00:00:00 2001
From: David Lougheed <david.lougheed@gmail.com>
Date: Sun, 14 Jan 2024 18:24:43 -0500
Subject: [PATCH] chore: configurable vcf chr transform fn

---
 config.example.js           | 10 ++++++++++
 epivar-prod/node1/config.js |  8 ++++++++
 epivar-prod/node2/config.js |  5 +++++
 models/samples.mjs          |  4 ++--
 models/tracks.mjs           |  2 +-
 5 files changed, 26 insertions(+), 3 deletions(-)

diff --git a/config.example.js b/config.example.js
index 5d6a85e2..de1497ea 100644
--- a/config.example.js
+++ b/config.example.js
@@ -31,5 +31,15 @@ module.exports = {
      * the metadata.json "donor" property).
      */
     vcfSampleNameConverter: name => name.split('_')[1],  // name => name
+
+    /*
+     * When given a contig from a peak as input, produce a contig compatible with the genotype VCF.
+     * For example, peak contigs may be in the format chr1, chr2, ..., whereas VCF contigs may be
+     * formatted without the 'chr' prefix (1, 2, ...). Stripping the chr prefix gives the correct
+     * VCF contig value.
+     */
+    vcfChrTransform: (chr) => chr.replace(/^chr/, ""),  // e.g., chr1 => 1
+
+    // bigWigChrTransform: TODO,
   },
 };
diff --git a/epivar-prod/node1/config.js b/epivar-prod/node1/config.js
index 5d6a85e2..3d1ea207 100644
--- a/epivar-prod/node1/config.js
+++ b/epivar-prod/node1/config.js
@@ -31,5 +31,13 @@ module.exports = {
      * the metadata.json "donor" property).
      */
     vcfSampleNameConverter: name => name.split('_')[1],  // name => name
+
+    /*
+     * When given a contig from a peak as input, produce a contig compatible with the genotype VCF.
+     * For example, peak contigs may be in the format chr1, chr2, ..., whereas VCF contigs may be
+     * formatted without the 'chr' prefix (1, 2, ...). Stripping the chr prefix gives the correct
+     * VCF contig value.
+     */
+    vcfChrTransform: (chr) => chr.replace(/^chr/, ""),  // e.g., chr1 => 1
   },
 };
diff --git a/epivar-prod/node2/config.js b/epivar-prod/node2/config.js
index 9d7c78c6..88731f54 100644
--- a/epivar-prod/node2/config.js
+++ b/epivar-prod/node2/config.js
@@ -31,5 +31,10 @@ module.exports = {
      * the metadata.json "donor" property).
      */
     vcfSampleNameConverter: name => name.split('_')[1],  // name => name
+
+    /*
+     * VCF contigs have correct format, use identity function (default value)
+     */
+    // vcfChrTransform: (chr) => chr,
   },
 };
diff --git a/models/samples.mjs b/models/samples.mjs
index d7b9ddcf..d1d49ef5 100644
--- a/models/samples.mjs
+++ b/models/samples.mjs
@@ -13,7 +13,6 @@ import {
   GENOTYPE_STATE_HET,
   GENOTYPE_STATE_HOM,
   GENOTYPE_STATE_REF,
-  normalizeChrom,
 } from "../helpers/genome.mjs";
 
 const { TabixIndexedFile } = Tabix;
@@ -22,13 +21,14 @@ const VCF_TABIX_FILE = new TabixIndexedFile({ path: envConfig.GENOTYPE_VCF_PATH
 const vcfParser = new VCF.default({ header: await VCF_TABIX_FILE.getHeader() });
 const vcfFilterFn = config.samples?.vcfFindFn
   ?? ((line) => line.REF.length === 1 && line.ALT.every((a) => a.length === 1));
+const vcfChrTransform = config.samples?.vcfChrTransform ?? ((chr) => chr);
 
 export default {
   queryMap,
 };
 
 export function queryMap(chrom, start, end = start + 1) {
-  return vcfQuery(normalizeChrom(chrom), parseInt(start.toString(), 10), parseInt(end.toString(), 10))
+  return vcfQuery(vcfChrTransform(chrom), parseInt(start.toString(), 10), parseInt(end.toString(), 10))
     .then(normalizeSamplesMap);
 }
 
diff --git a/models/tracks.mjs b/models/tracks.mjs
index 29c42366..f8abf388 100644
--- a/models/tracks.mjs
+++ b/models/tracks.mjs
@@ -53,7 +53,7 @@ function get(peak) {
 
 async function values(peak, usePrecomputed = false) {
   const k = `values:${peak.id}${usePrecomputed ? ':pre' : ''}`;
-  const chrom = normalizeChrom(peak.feature.chrom);
+  const chrom = normalizeChrom(peak.feature.chrom);  // TODO: bigWigChrTransform
 
   await cache.open();