Error reading full chunk

[cbossu]

Hello,

I'm running into an error while formatting the glf file to run ngsDist.

-> Printing at chr: scaffold1|size15271748 pos:3044951 chunknumber 60900 contains 50 sites
-> Printing at chr: scaffold1|size15271748 pos:3049951 chunknumber 61000 contains 50 sites
-> Printing at chr: scaffold1|size15271748 pos:3054951 chunknumber 61100 contains 50 sites
-> Printing at chr: scaffold1|size15271748 pos:3059951 chunknumber 61200 contains 50 sites

[glfReader.cpp:fetch():34] Error reading full chunk: bytesRead:3157 expected:13600 will exit

I don't think it is a memory issue as it stops at this point when I run in on a regular memory and large memory node. I originally thought it was a glf issue (so I ran glf in two ways --doGlf 2 and --doGlf 4), but I get the same error. There was a previous post about this that said the glf might be corrupt, however, I can use the glf in ngsLD without an error. Do you have any ideas why this error is being thrown? Thanks.

CH

[fgvieira]

Is the number of sites correct? Which command line are you using?

[cbossu]

I used the following code:

Text glf with normal scale

$NGSTOOLS/angsd/angsd -P 16 -glf $outfile.glf.gz -fai BUOW.fasta.fai -nInd 170 -doMajorMinor 1 -doPost 1 -doMaf 1 -doGeno 8 -out all_incl_SC.rmrelFinal

So I haven't even run ngsDist yet, where the nsites actually comes into play. I'm just preparing the genotype file. I used the glf 4 option. Thanks for your help.

Ch

[fgvieira]

But then why are you using GL files as input? I'd suggest you try to generate it directly from the BAM/CRAM files.

However, it looks more like an angsd issue, and it might be better to raise the issue on the angsd github.