Reads with / characters in the read names cause the pipeline to fail

[marc-jones]

Check Documentation

I have checked the following places for your error:

• nf-core website: troubleshooting
• nf-core/clipseq pipeline documentation

Description of the bug

If / characters are in the read names, then the pipeline will fail at the dedup step. This is caused, however, by STAR truncating read names based on the / character. This removes the UMI information which was added earlier in the pipeline, causing the dedup step to fail.

Steps to reproduce

Steps to reproduce the behaviour:

Running the pipeline with reads with a / in the read name. Example

Expected behaviour

Either a verbose error message saying that the reads are incompatible, or automatic removal or replacement of the / characters.

[marc-jones]

Interestingly, when there's a space in the read it works correctly:

# initial format
@ERR039787.1 HWI-EAS202_223:6:1:37:2026/1

# format after move_umi
@ERR039787.2_GTTCCAC HWI-EAS202_223:6:1:41:30/1

# format after STAR mapping
ERR039788.1477620_GAAGCCC