Improve/add UMI deduplication metrics

[ppericard]

Description of feature

Hello ^^
I'm having difficulties finding easy to understand stats on UMI deduplication in the outputs. It seems there is no section in the multiqc output, not even in the statistics table (where I would expect to have metrics about nb of reads before dedup, nb of reads after dedup, % duplication (from umi-tools dedup on alignements)).
In the output directory, I'm also not finding a log with easy to understand metrics from umi-tools dedup. I'm probably missing something.
Thanks in advance. Pierre

[MatthiasZepper]

Since people complained about the poor performance, the generation of deduplication statistics if off by default now.

You have to set the parameter --umitools_dedup_stats respectively umitools_dedup_stats : true in a params file to activate that functionality.

[ppericard]

Hi @MatthiasZepper,
I'm sorry if i wasn't clear enough in my initial message. All my comments apply to the pipeline while having activated the --umitools_dedup_stats parameter.
In the *.umi_dedup.transcriptome.filtered.prepare_for_rsem.log files there are no summaries with the dedup stats, and the other files are not very informative and easy to read: *.umi_dedup.sorted_edit_distance.tsv, *.umi_dedup.sorted_per_umi_per_position.tsv, *.umi_dedup.sorted_per_umi.tsv. There is a real need for an easy to read and understand summary for deduplications, such as the one that can be obtained through Multiqc parsing of the UMI tools for exemple (MultiQC/MultiQC#1769).
Right now, as a user I have even less information about deduplication than what I would have in the logs just by running the umi-tools dedup command.

[MatthiasZepper]

Apologies for stonewalling on this issue before. While hunting down the cause for issue #1303, it occurred to me that probably a botched MultiQC config is behind this issue as well. For some reason, we explicitly specify the MultiQC modules to be run and UMI-tools is nowhere to be found.

Since we run MultiQC with a custom config outside the pipeline again, we did not notice.

It should be fixed on this branch, but I struggle with testing at the moment.