Negative coverage depths when running jgi_summarize_bam_contig_depths (Metabat2)

Question

[amir.rakhimov.b]

Hello,

I was assembling metagenomic reads into contigs and binning contigs with MegaHIT and MetaBAT2, and noticed that the depth file from jgi_summarize_bam_contig_depths contained negative depths in the totalAvgDepth column. So when I ran metabat2 command, I got a warning:

[Warning!] Negative coverage depth is not allowed for the contig k141_141795, column 1: -1.46002e+09 
[Warning!] Negative coverage depth is not allowed for the contig k141_171645, column 1: -6.6807e+08

All of my samples contained 2-3 such contigs. Additionally, the jgi_summarize_bam_contig_depths showed this warning:

WARNING: your aligner reports an incorrect NM field.  You should run samtools calmd! nm < ins + del: cmatch=0 nm=13 ( insert=0 + del=55 + mismatch=13 == 68) A00609:874:HC557DSX7:3:2371:5864:9079

I thought that the incorrect NM field could have caused the depths to be negative. But when I ran

samtools calmd -b sample1_bothReadsMapped_sorted.bam  sample1_final.contigs.fa \
 > sample1_mapped_and_unmapped_fixed.bam

I kept getting such errors for every contig:

[bam_fillmd] fail to find sequence 'k141_231346 flag=0 multi=126.6091 len=1937' in the reference.
[W::fai_get_val] Reference k141_231346 flag=0 multi=126.6091 len=1937 not found in FASTA file, returning empty sequence

The question is: How can the negative depths and the NM field be fixed? I would appreciate your help.

My command for Metabat2 (it was a for loop on many samples, but this is the basic example) was:

metabat2 -i sample1.final.contigs.fa \
    -a depth.txt  \
    -o sample1_bins/sample1_bin \
    -v

And the jgi_summarize_bam_contig_depths command (also a for loop) was:

jgi_summarize_bam_contig_depths --percentIdentity 97 \
    --minContigLength 1000 \
    --minContigDepth 1.0  \
    --referenceFasta sample1_final.contigs.fa \
    sample1_bothReadsMapped_sorted.bam \
    --outputDepth sample1.depth.txt

The BAM file was sorted (I checked with samtools view -H sample1_bothReadsMapped_sorted.bam |grep "SO:coordinate" and got only @HD VN:1.4 SO:coordinate).

I am using metabat2 with conda on Windows subsystem for Linux.

Answer 1

[guangingmai]

Please install newest metabat2 in the following url: https://bitbucket.org/berkeleylab/metabat/src/master/. The metabat2 package in conda is not supported by the author. You can search the same issus in the issus tracker.

[amir.rakhimov.b]

Thank you very much! I built a docker image, and jgi_summarize_bam_contig_depths created depth files without negative values.