search

Arkadiy-Garber / MagicLamp

A platform for targeted annotation of (meta)genomic and (meta)transcriptomic datasets using HMM sets.
GNU General Public License v3.0
13 stars 5 forks source link

Different results of LithoGenie and FeGenie #7

Open kodragonP opened 2 years ago

kodragonP commented 2 years ago

I Found that LithoGenie can predict more cyc2 gene than FeGenie. Why? Which one I can believe? Moreover, I found the results of FeGenie were very unstable. I can predict a cyc2_cluster3 from a genome when together with other 10 genomes. However, if I only predict from this only one genome, no cyc2 was reported.

seanmcallister commented 1 year ago

Arkadiy: Have you looked into this issue with the clustering?

kodragon: The cluster3 hmm of the cyc2 trio is probably the least stable of the three. That particular cluster of the gene tree has lots of deep branches and is not highly supported. This might be one reason it isn't consistent. Likely, the clustering is picking the longest representative, and that one is hitting above the bit score cutoff. When you run the genome individually that was chosen as the representative for the gene cluster, does it pick up the cyc2 cluster 3?

Arkadiy-Garber commented 1 year ago

I haven't yet looked into this. I haven't been able to replicate this error on my end.

@kodragonP - iron HMMs from LithoGenie have been deprecated and will be removed shortly. Please use FeGenie for iron gene prediction.

Thanks, Arkadiy