From 364dc6598e11151fd0b146814b2aaac272eec8d7 Mon Sep 17 00:00:00 2001
From: Frederick Muriuki Muriithi
Date: Mon, 5 Feb 2024 15:42:16 +0300
Subject: Run QC on the 'geno' file in R/qtl2 bundle.

---
 scripts/process_rqtl2_bundle.py | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

(limited to 'scripts')

diff --git a/scripts/process_rqtl2_bundle.py b/scripts/process_rqtl2_bundle.py
index 6ba746a..4da3936 100644
--- a/scripts/process_rqtl2_bundle.py
+++ b/scripts/process_rqtl2_bundle.py
@@ -11,6 +11,10 @@ from logging import Logger, getLogger, StreamHandler
 import MySQLdb as mdb
 from redis import Redis
 
+from functional_tools import take
+
+import r_qtl.errors as rqe
+import r_qtl.r_qtl2 as rqtl2
 import r_qtl.r_qtl2_qc as rqc
 
 from qc_app import jobs
@@ -60,6 +64,19 @@ def qc_r_qtl2_bundle(bundlefilepath, logger):
         logger.info(
             "Bundle successfully validated. All listed files are present.")
 
+        cdata = rqtl2.control_data(zfile)
+        if "geno" in cdata:
+            gerr = False
+            logger.info("Validating 'geno' file.")
+            for error in take(rqc.geno_errors(zfile), 10):
+                gerr = True
+                logger.error("%s: [Line %s, Field %s]",
+                             error[2], error[0], error[1])
+            if gerr:
+                logger.error("... more")
+                raise rqe.InvalidFormat("'geno' file content contains errors.")
+            logger.info("'geno' file validation was successful.")
+
 def process_bundle(dbconn: mdb.Connection,
                    rconn: Redis,
                    rprefix: str,
-- 
cgit v1.2.3