From a5def929133c457f42ddd17e6399f11083114889 Mon Sep 17 00:00:00 2001
From: Jonas Fuchs <78491186+jonas-fuchs@users.noreply.github.com>
Date: Fri, 1 Dec 2023 12:46:49 +0100
Subject: [PATCH] introduced -q arg (#13)

---
 bamdash/command.py      | 10 +++++++++-
 bamdash/scripts/data.py |  4 ++--
 2 files changed, 11 insertions(+), 3 deletions(-)

diff --git a/bamdash/command.py b/bamdash/command.py
index 38a323b..a259029 100644
--- a/bamdash/command.py
+++ b/bamdash/command.py
@@ -44,6 +44,14 @@ def get_args(sysargs):
         metavar=" ",
         help="seq reference id"
     )
+    parser.add_argument(
+        "-q",
+        "--quality-threshold",
+        type=int,
+        default=15,
+        metavar="15",
+        help="qaulity threshold for reads"
+    )
     parser.add_argument(
         "-bs",
         "--binsize",
@@ -121,7 +129,7 @@ def main(sysargs=sys.argv[1:]):
     args = get_args(sysargs)
 
     # define subplot number, track heights and parse data
-    coverage_df, title = data.bam_to_coverage_df(args.bam, args.reference, args.coverage)
+    coverage_df, title = data.bam_to_coverage_df(args.bam, args.reference, args.coverage, args.quality_threshold)
     track_heights = [1]
     track_data = []
     # extract data and check if ref was found
diff --git a/bamdash/scripts/data.py b/bamdash/scripts/data.py
index 054f059..7712e34 100644
--- a/bamdash/scripts/data.py
+++ b/bamdash/scripts/data.py
@@ -63,7 +63,7 @@ def make_title_string(parsed_bam, coverage_df, reference, min_cov):
     return stat_string
 
 
-def bam_to_coverage_df(bam_file, ref, min_cov):
+def bam_to_coverage_df(bam_file, ref, min_cov, quality_thres):
     """
     :param bam_file: bam location
     :param ref: chrom identifier
@@ -75,7 +75,7 @@ def bam_to_coverage_df(bam_file, ref, min_cov):
 
     coverage, position = [], []
     # count coverage at each pos
-    coverage_base = bam.count_coverage(ref)
+    coverage_base = bam.count_coverage(ref, quality_threshold=quality_thres)
     # extract overall coverage and pos
     for index, [A_count, C_count, G_count, T_count] in enumerate(
             zip(coverage_base[0], coverage_base[1], coverage_base[2], coverage_base[3])):