[FEAT] Domain architecture analyses

[jananiravi]

Clean implementation of RPSBLAST w/ our existing DA analyses

rps analysis

• rpsblast using CDD
  • make sure the output is in A+B+C for each of the accessions
  • + slurm
  • append final DA column to BLAST output

rps2da

• fix domain addition
• add clean COG db to lookup file (if not already present)

rps2viz

• create, similar to ipr2viz

Alternatives

functionality and final implementation should be similar to RPSBLAST

• HMMer | HMMer-Userguide.pdf
• individual runs? | ask @jananiravi before working on this
  • SignalP | signalp | manual
  • TMHMM | tmhmm | manual
  • + slurm

[jakekrol]

@jananiravi @epbrenner, do we want to summarize how exactly we'd like RPSBLAST implemented?

Can we identify where in the pipeline of upstream_scripts this will go?

Atm, it's towards the end of full analysis for homologs (i.e., does not run for queries): https://github.com/JRaviLab/molevol_scripts/blob/main/upstream_scripts/00_wrapper_full.sb#L152-L155

However, I don't see any rps output files in the job results folders 🤔

[jananiravi]

That spot seems to be correct. As shown here, 04a (ipr) and 04b (rps) were meant to be alternate ways of going about prot2da. There's a good amount of overlap, but there are also rps/cdd-specific vs. ipr-specific analyses. In this case, we could potentially integrate the best of both worlds.

I know @d33bs and @epbrenner were looking into hmmer as well. These will all help us go from prot2da (proteins to domains/domain_architectures).

No temp files from rps yet since 04b was never really used to generate these files. Once we run that (w/ blast+ suite, we might have an answer to that).

cc: @the-mayer