Enhancement: add whitelist to pre-processing

[dhoconno]

Thanks for this very useful workflow. To reduce runtimes, can I suggest adding a 'whitelist' rule to preprocessing? This could reduce runtimes considerably in situations where the targets are limited (e.g., only interested in known human viruses).

I think the implementation could be straightforward:

• Add a config option to specify a FASTA/Q file of sequences to whitelist.
• In the preprocessing rule files, duplicate the existing host_removal_mapping rule to whitelist_read_mapping or equivalent
• Instead of excluding mapped reads with samtools view -f 4..., the duplicated rule would map reads to the whitelist and retain only mapped reads with samtools view -F 4.

Thanks for your consideration!

[beardymcjohnface]

Hi, This is an interesting suggestion. Do you think having an option to use a custom primary database for the viruses of interest would work? The primary searches do essentially what you're suggesting, but for all viruses, and the secondary multi-kingdom searches weed out the false positives from this reduced pool of sequences.

[dhoconno]

Yep, absolutely. Depending on how much database prep is needed for the primary database, I could envision situations where providing a FASTA whitelist file would be simpler and wouldn't require modifying the virus database. If the primary database is already just a FASTA file of all viruses, then specifying a custom FASTA file of, say, all human viruses would be great.