Some questions regarding data prep?

[bananabenana]

Hi,

Thanks for developing deltaBS. I am having a bit of trouble prepping my data and was hoping you could clarify a few things.

The background: I have 8k Klebsiella genomes I want to generate DBS values for.

• I have been running hmmsearch using Gammaproteobacteria eggnog HMM profiles, but this is taking a very long time. Do you have any recommendations as to the most appropriate way to generate HMM bitscores in a relatively short period of time?
• Is there any reason I can't use hmmscan over hmmsearch to generate the bit scores? Just that it will be quicker for me to screen genes against HMMs rather than HMMs against genes.
• As I'm using a Panaroo input as per the 'deltaBS across a population of bacteria' wiki page, would it be better for me to build my own HMMs of each gene, and is this the buildCustomModels.pl script?
• Furthermore on this, could I use my pangenome protein fasta file as --database and for the --proteome option?

Thanks

[nwheeler443]

Hi,

Thanks for using deltaBS!

Re: generating the HMM bitscore data quickly, the best approach we've got for running many genomes is just to run the search on each proteome file individually and submit many parallel jobs - it can still be pretty time consuming but this is a general issue with analysing all genes in a large collection of a species with a big pangenome. There's no meaningful difference results-wise between using hmmsearch and hmmscan, so if your genes:models ratio favours the reverse, go for it

Re: choice of HMMs, there are pros and cons to using EggNOG vs custom models - custom models will cover more of your pangenome, but are time consuming in themselves to build. EggNOG models have less coverage but are built using a more sophisticated and curated process, so can potentially be higher quality. If you do want to build your own reference HMM database, using the reference sequence for each gene from Panaroo would be ideal, and yes, you've identified the right script.