[CHERRY] TopN prediction instead of Top1

[valentynbez]

Hello,

In CHERRY publication in FIg. 10 it shows, that predicting topN hosts improves accuracy of CHERRY. However now it outputs only top 1 prediction. Is there a way to run a software with topN outputs?

Cheers,
Valentyn

[KennthShang]

Hi @valentynbez ,

Several reasons for this:

1. We found that most of the databases only report one host of a virus. So, it seems that returning topk is not a practical choice. We also noticed that feedback from PhaBOX users revealed some parameters no one used. So we remove them in the updates.
2. The program is slightly different from the original program reported in the paper (for higher speed and more precise for the metagenomic sequencing data). We will clarify this in our latest paper.
3. If you are using MAG mode for host prediction, you can use the CRISPRs_alignment_MAG.tsv in the cherry_supplementary folder. This file contains all the candidate hoststhat meet the threshold. There is also another file named crispr_out.tab in the midfolder with header qseqid sseqid evalue pident length slen. This is all the alignment results between MAG's CRISPR and input viruses.

Best,
Jiayu