pipeline fails on qnorm step when replicate names contain a common prefix or postfix

[Thomas Gilgenast]

The qnorm step of the pipeline will break if the rep names contain a common prefix or postfix (e.g., "ES_rep1" and "ES_rep2").

This is because these replicate names will get mangled by lib5c.tools.helpers.infer_replicate_names() to substrings of their full name (e.g., "1" and "2") when the lib5c qnorm command is deciding where to write output. This leaves files on disk like "1_qnorm.counts", but the pipeline expects output files exactly equal to outfile_pattern % rep_name, or in this example "ES_rep1_qnorm.counts". The pipeline then assumes the qnorm step failed since the expected output files are missing (in reality they were generated successfully but have unexpected filenames).

The short-term workaround is that replicate names with common prefix or postfix are not supported.

[Thomas Gilgenast]

see #41 for a proposed solution

[Thomas Gilgenast]

fixes #39, #40

closes #41 (it uses the approach proposed there)

→ <<cset 1ec4da7073eb>>