In the documentation provided here: https://nf-co.re/smrnaseq/2.4.1/docs/usage/ , you mention in a warning after the UMI handling section that the execution order is fastp → umicollapse → umi_tools extract. Therefore, if you are using the QIAseq kit with the adapter/anchor sequence positioned before the UMI, it could get deleted by fastp, changing the read structure and therefore umi_tools extract will no longer being able to grab the UMI sequence.
However, in the Output section, it appears that the order is actually umi_tools extract -> umicollapse -> fastp, and this makes far more sense, as we would like to get rid of very short reads after getting rid of the UMI and adapters from the read. Moreover, the QIAseq adapter before the UMI wont get removed this way.
What is the actual execution order?
In the documentation provided here: https://nf-co.re/smrnaseq/2.4.1/docs/usage/ , you mention in a warning after the UMI handling section that the execution order is
fastp→umicollapse→umi_tools extract. Therefore, if you are using the QIAseq kit with the adapter/anchor sequence positioned before the UMI, it could get deleted byfastp, changing the read structure and thereforeumi_tools extractwill no longer being able to grab the UMI sequence.However, in the Output section, it appears that the order is actually
umi_tools extract->umicollapse->fastp, and this makes far more sense, as we would like to get rid of very short reads after getting rid of the UMI and adapters from the read. Moreover, the QIAseq adapter before the UMI wont get removed this way.What is the actual execution order?