From 63669d435e5a8c3b2ca36ef89274c4d19dd4e34a Mon Sep 17 00:00:00 2001
From: Holly Ruess <s185797@biohpcwsc012.biohpc.swmed.edu>
Date: Tue, 2 Jul 2019 16:18:19 -0500
Subject: [PATCH] fix issue 58 single rep
---
workflow/scripts/pool_and_psuedoreplicate.py | 49 ++++++++++----------
1 file changed, 25 insertions(+), 24 deletions(-)
diff --git a/workflow/scripts/pool_and_psuedoreplicate.py b/workflow/scripts/pool_and_psuedoreplicate.py
index f6ba5a8..eb7c76d 100644
--- a/workflow/scripts/pool_and_psuedoreplicate.py
+++ b/workflow/scripts/pool_and_psuedoreplicate.py
@@ -301,31 +301,32 @@ def main():
# if so replace with pool_control
# unless single control was used
- if not single_control:
- path_to_pool_control = cwd + '/' + pool_control
- if control_df.values.max() > cutoff_ratio:
- logger.info("Number of reads in controls differ by " +
- " > factor of %f. Using pooled controls." % (cutoff_ratio))
- design_new_df['control_tag_align'] = path_to_pool_control
- else:
- for index, row in design_new_df.iterrows():
- exp_no_reads = utils.count_lines(row['tag_align'])
- con_no_reads = utils.count_lines(row['control_tag_align'])
- if con_no_reads < exp_no_reads:
- logger.info("Fewer reads in control than experiment." +
- "Using pooled controls for replicate %s."
- % row['replicate'])
+
+ if not single_control:
+ path_to_pool_control = cwd + '/' + pool_control
+ if control_df.values.max() > cutoff_ratio:
+ logger.info("Number of reads in controls differ by " +
+ " > factor of %f. Using pooled controls." % (cutoff_ratio))
+ design_new_df['control_tag_align'] = path_to_pool_control
+ else:
+ for index, row in design_new_df.iterrows():
+ exp_no_reads = utils.count_lines(row['tag_align'])
+ con_no_reads = utils.count_lines(row['control_tag_align'])
+ if con_no_reads < exp_no_reads:
+ logger.info("Fewer reads in control than experiment." +
+ "Using pooled controls for replicate %s."
+ % row['replicate'])
+ design_new_df.loc[index, 'control_tag_align'] = \
+ path_to_pool_control
+ else:
+ if paired:
+ control = row['control_tag_align']
+ control_basename = os.path.basename(
+ utils.strip_extensions(control, STRIP_EXTENSIONS))
+ control_tmp = bedpe_to_tagalign(control, control_basename)
+ path_to_control = cwd + '/' + control_tmp
design_new_df.loc[index, 'control_tag_align'] = \
- path_to_pool_control
- else:
- if paired:
- control = row['control_tag_align']
- control_basename = os.path.basename(
- utils.strip_extensions(control, STRIP_EXTENSIONS))
- control_tmp = bedpe_to_tagalign(control, control_basename)
- path_to_control = cwd + '/' + control_tmp
- design_new_df.loc[index, 'control_tag_align'] = \
- path_to_control
+ path_to_control
else:
path_to_pool_control = cwd + '/' + pool_control
--
GitLab