nah, 'cause it'll be me, too!

well, if your preferred tipple of choice is snakemake, maybe you can use whatever it is I build 😭

yep yep. curious if you know of a package or workflow to do this, from e.g. blastx or diamond? I can dream something up easily enough but I'm already about 12 workflows in on this paper and getting a bit tired of double-cross-checking all the code 😅.

yes, exactly - we are basically at that point of saying, well, it doesn't classify so... can we assume it's not euk? :)

yeah, that's pretty much where we are - gene calls and sequence gazing. Hard to do on a large scale though, or at least it feels annoying to me.

with BLAST, translated.

problem is that you can only do this on marker genes. So we know the MAG itself is "good", by usual standards. But want to rule out (as best possible) contigs that are unknown contam.

virus would actually be fine ;). Just want to avoid euk.

(if you need names, jill banfield and meren would be the people I'd trust more than most anyone else ;))

I will put you on the "preprint available" list and wait for your shots on target :)

Also, I know too much about assembly to trust contigs past a certain point :)

we have "good" MAGs. We have reads matching to the MAGs from metagenomes. So we could do contig and read analysis both, but I feel like the reads are the primary data. Assembly is possible but not in all cases. Trying to rule out (as best possible) euk matches, esp from host.

thx!