[QUESTION] Global coverage calculation

[ziphra]

Hello,

I hope you are well

I noticed that the global coverage reported by Robin seems much lower than the value I obtain with mosdepth. For example, with 25 GB of data, I expect around 6.6X, which is also approximately what mosdepth reported for my sequencing runs of this size. But in the Robin report I had 0.92X and 2.79X. Do you have an idea why there is this discrepancy?
Could you clarify the contents and calculations behind coverage_main.csv and bed_coverage_main.csv? The coverage and meandepth fields here do not appear accurate on my side, whereas the coverage column in target_coverage.csv does look correct.

Could all this be related to the fact that i hard coded my bed file to be used for SNV calling and for target coverage calculation ?

Many thanks in advance

Euphrasie

[mattloose]

Which mosdepth command are you using and what input are you giving it?

If you are modifying the bedfiles then the system may not perform as expected. We have a version of robin coming that allows for user specified bed files.

[ziphra]

Thank you for your response, this is the command i use :

mosdepth -t 30 --by genes.bed out sample.bam

And in the summary.txt, i will have the mean coverage on my targets and a global mean depth. The targets mean coverage match what is reported by Robin.

[mattloose]

OK - I shall double check what is happening here with some reference data.

[ziphra]

just to be sure, I tested without my bed, using the default settings, and I still didnt get the expected global coverage, even though i had a different result.

With my bed:

Coverage Summary
Coverage Depths - Global Estimated Coverage: 0.92x Targets Estimated Coverage: 72.96x
Outlier genes by coverage (high):
Outlier genes by coverage (low): CDKN2A (39.1) - B2M (34.5) - BTK (32.3) - CDKN2B (21.7) - TNFRSF14 (19.2) - TNFRSF14-AS1 (17.4) - ID3 (16.5)

Default:

Coverage Depths - Global Estimated Coverage: 2.63x Targets Estimated Coverage: 10.32x
Outlier genes by coverage (high): EZH2 (81.7) - NOTCH2 (75.5) - CASC11,MYC (74.1) - MIR4673,NOTCH1 (73.0) - BRAF (73.0) - CREBBP (71.6) - ARID1A
(69.4) - NAB2,STAT6 (68.7) - TP53 (55.2) - CDKN2A,CDKN2B,CDKN2B-AS1 (32.4) - GNAQ (13.1) - LINC02511 (11.8) - FERMT1 (11.5) - SUMF1 (11.0) -
PLCH1 (10.5) - SNTB1 (10.4)
Outlier genes by coverage (low): PDYN,PDYN-AS1 (9.8) - PDP1 (9.7) - ADGRL2 (9.5) - NMD3 (9.4) - IL12RB2 (9.4) - CRMP1,EVC (9.3) - LCOR (3.6) -
EPS8L3,GSTM5 (3.5) - MIR4733HG,NF1 (3.2) - EZHIP (3.2) - ATRX (3.1) - STAG2 (3.1) - CTDNEP1 (3.0) - KDM6A (3.0) - SMARCB1 (2.9) - BCOR (2.8) -
DDX3X,RN7SL15P (2.8) - EIF1AX (2.7) - ABL1 (2.7) - FOXO3 (2.5) - KBTBD4,NDUFS3,PTPMT1 (2.4) - AXDND1,NPHS2 (2.3) - FOXR2 (2.1) - HRAS (2.1) -
WNK1 (2.1) - VCAN,VCAN-AS1 (2.0) - L2HGDH (2.0) - TTC3 (2.0) - TMEM266 (1.9) - DNMT1 (1.8) - ECE1 (1.8) - ABCB11 (1.2) - DAXX (1.2) - EDA (1.2) -
DDR1,MIR4640 (1.0)

I expect : global X: 5.95, ROIs X: 71.62

[mattloose]

Yeah - the default will definitely be incorrect because of the underlying assumptions mad by robin. I don't quite understand the global coverage being off.