metabat2 needs depth.txt?

[Weronika Jaśkowiak]

Hello,
I am trying to perform metabat analysis. I used command metabat2 -i assembly_from_medaka.fasta -a depth.txt -o output_directory. Unfortunately, when I provide depth.txt it doesn’t produce any bins, but without depth.txt it works. Is depth file important to obtain correct results? The same issue is, when I use runMetaBat.sh. Check the attached screenshot. Thanks in advance.

[Rob Egan]

Hello.

If you give metabat2 an abundance depth file, and it looks like you gave it an empty one, then only those contigs that have a sufficient minimum depth will be used. Since there was nothing in that file, all the contigs were screened out because none had that minimum depth (1.0 by default).

It is not strictly necessary to give metabat2 an abundance file, but it is highly recommended as that is a crucial part of the data to decide which contigs are from the same genome. If you leave out the empty depth file (i.e. no “-a depth.txt” in the command line), then metabat will attempt to bin on just the TNF frequencies alone.

Also, as your file, assembly_from_medaka.fasta, implies, your mileage may vary if the assembly has any indels or imprecise mapping from ONT based assemblies. MetaBAT2 has not been well tested on such assemblies.

-Rob

[Rob Egan]

Correction. After looking at your logs more closely and the code, it looks like there were 8107 large and 12 small contigs in your assembly, but only 137 contigs in the depth file which you provided (it wasn’t actually empty). But, why do so few contigs have an entry in the depths file?

I just pushed a new version that should abort with a better message in this case where there are no large contigs with sufficient depth.

You can adjust the default minimum depths with the --minCV (1.0 default) and --minSumCV (1.0 default) options. Note that the sum of the depths for this filter only counts those contig-samples which are each > minCV depth.

[Weronika Jaśkowiak]

Thank you for the explanation! I truly appreciate it.