search

Phelimb / atlas

atlas
MIT License
5 stars 4 forks source link

Issues from Rachel #10

Open Phelimb opened 8 years ago

Phelimb commented 8 years ago

Right, so it works amazingly well except in a couple of cases, and Zam told me to mention them to you/raise a bug for them with you. I didn’t do much kmer size fiddling, and I ended up sticking with the overestimate of covg size for nanopore, which might have confused some things.

TEM-1 vs TEM-40: Still guessing TEM-40 in most cases, which is much less likely than TEM-1, 1 Amino Acid away. In one case I have Atlas guessing TEM-1 for PacBio assembly and TEM-40 for Nanopore. In several other cases it guesses TEM-40 for both.

  1. KPC-2 vs KPC-5: I have an example where Atlas correctly guesses KPC-2 at lower covg for a nanopore sample (<= 5898 reads), but switches to guessing KPC-5 at higher covg (>=6059 reads). This was however for the sample done with the oldest Nanopore chemistry, so the per base accuracy will be lower in the reads.

I’m not sure what info you’d like on these, but I can give you them as examples.

Phelimb commented 8 years ago

@rmnorris

Thanks!

I'll take a look at these and let you know. Could you possibly send the sample ids of where you saw these?

rmcolq commented 8 years ago

CAV1741 had the TEM-1 in PacBio, TEM-40 in Nanopore situation, and CAV1016 had the KPC switching issue, P46212 guessed TEM-40 in both PacBio and Nanopore.

Full ids: JR_FAA63658_29092015_ecol_P46212, MN15229_FAA89259_09022016_cfre_CAV1741, JR_FAA35548_13042015_KpneCAV1016

On the fly working in directory /data2/users/rachel/projects/initial/ECCMID/on_the_fly/ with directories for each sample and subdirectories for numbers of reads including fasta.