CoverM and jgi_summarize_bam_contig_depths give different outputs

[apcamargo]

Hi Ben

I just replaced jgi_summarize_bam_contig_depths with CoverM in my binning pipeline (which uses both MetaBat2 and MaxBin2) and I've noticed that my final bins was different from the ones I got before.

I prepared reduced BAMs, with just three contigs, and measured coverage with both jgi_summarize_bam_contig_depths and CoverM:

contigName	method	contigLen	totalAvgDepth	BE_BS_R1-BE_BS_R1.reduced.bam	BE_BS_R1-BE_BS_R1.reduced.bam-var	BE_BS_R1-BE_RX_R1.reduced.bam	BE_BS_R1-BE_RX_R1.reduced.bam-var	BE_BS_R1-BM_ER_R7.reduced.bam	BE_BS_R1-BM_ER_R7.reduced.bam-var	BE_BS_R1-BM_RX_R7.reduced.bam	BE_BS_R1-BM_RX_R7.reduced.bam-var
BE_BS_R1_k147_2106787	CoverM	3152	5.5702863	5.578614	5.697416	14.260826	23.711023	0.92271817	2.084662	1.5189873	2.341692
BE_BS_R1_k147_2106787	jgi_summarize_bam_contig_depths	3152	22.2811	5.57861	5.69755	14.2608	23.7111	0.922718	2.08466	1.51899	2.34171
BE_BS_R1_k147_3889603	CoverM	2203	3.069654	10.337555	50.398186	0.71310276	1.1237903	0.49245006	0.889441	0.73550904	0.8008681
BE_BS_R1_k147_3889603	jgi_summarize_bam_contig_depths	2203	12.2786	10.3376	50.3978	0.713103	1.1238	0.49245	0.889457	0.735509	0.800879
BE_BS_R1_k147_2120400	CoverM	3251	4.63673	4.990003	5.4105444	8.594324	14.234088	0.65011287	1.901733	4.31248	36.406517
BE_BS_R1_k147_2120400	jgi_summarize_bam_contig_depths	3251	18.5469	4.99	5.41058	8.59432	14.2341	0.650113	1.90183	4.31248	36.405

reduced_bam.zip

As you can see, there are two differences:

1. The totalAvgDepth differs a lot between the two tools. CoverM got the values right and I don't know how jgi_summarize_bam_contig_depths computed its results.
2. CoverM outputs more decimal cases then jgi_summarize_bam_contig_depths. I think that's the reason the binning performed with CoverM gave me different results.

I'm wondering if CoverM should use the same number of decimal cases when using -m metabat output, just to keep consistency with jgi_summarize_bam_contig_depths.

Thank you for the great tool!

[wwood]

Hi,

Thanks for the kind words. jgi_summarize_bam_contig_depths has been updated from its original form e.g. https://bitbucket.org/berkeleylab/metabat/issues/48 - are you using a new version?

I think you are right about the decimal places - I've fixed this in ada7540 citing you.

[apcamargo]

I'm using the version in Bioconda. Probably it wasn't updated there.

Thank you for the prompt response!

[wwood]

HI,

OK, maybe that's it. If you have time I'd encourage you to update the bioconda repo and/or report your results here to the metabat peeps - You have a nice reduced BAM files which will make the bug easier to fix on their end, if it is still a problem. Thanks again for the report. Closing now - let me know if there's anything else. ben

On Sat, 8 Feb 2020 at 07:38, Antônio Pedro Camargo notifications@github.com wrote:

I'm using the version in Bioconda. Probably it wasn't updated there.

Thank you for the prompt response!

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/wwood/CoverM/issues/19?email_source=notifications&email_token=AAADX5DHC3BCUJZOWTWVVVDRBXIHBA5CNFSM4KRDONI2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOELEWZNY#issuecomment-583625911, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAADX5GFE4Z2CTAOATXMUXDRBXIHBANCNFSM4KRDONIQ .

--

Ben Woodcroft http://ecogenomic.org/users/ben-woodcroft http://www.ecogenomic.org/

[Tarasovk49]

Hi! CheckM produces the mean of four values and jgi_summarize_bam_contig_depths produces the sum. You may easily check it: 22.2811 / 4 = 5.70275; 12.2786 / 4 = 3.06965; and so on.. The values you got with CheckM. Inconsistencies in last cases are due to the reason you mentioned: CheckM store values with more decimal places. Hope that helps!