"Core Dumped" when using GEMMA for GWAS

Question

[amandinette]

Dear all,

I am following this tutorial about bacterial GWAS using GEMMA and Bugwas : https://github.com/janepipistrelle/bacterial_GWAS_tutorial/blob/master/tutorial.rmd

When I try correct for population structure, I obtain a segmentation fault.

Rscript bacterialGWAS/kmerGWAS/kmerLMM.R -chisqStat fus300.gwaskmer-out.chisqStat.txt -patternKey fus300.gwaskmer-out.patternKey.txt -patternIndex fus300.gwaskmer-out.patternIndex.txt -signif 5000 -relateMatrix ~/data/fus300_gemma_relmatrixout.cXX.txt -phenotype data/fus300_bugwas.pheno.txt -prefix fus300 -externalSoftware dependencies.txt

Running kmerLMM.R (...) Produce a matrix to fill to create the GEMMA gen file: Create GEMMA snp file: # Find the patterns which each kmer corresponds to out of those used in the LMM: (...) Reading Files ... Segmentation fault (core dumped) Error in file(file, "rt") : impossible d'ouvrir la connexion Calls: read.table -> file De plus : Warning message: In file(file, "rt") : impossible d'ouvrir le fichier './output/fus300_kmer_lmmout.assoc.txt' : Aucun fichier ou dossier de ce type Exécution arrêtée

I can't understand why... Can someone help me ?

Thank you !

Amandine

Answer 1

[rnnh]

"Segmentation fault" means that you tried to access memory that you do not have access to. You need more memory to run your job.

Try requesting more memory using qsh, qlogin or qrsh. For example, the following command will request "X" MB:

$ qlogin -l h_vmem=XM

Answer 2

[morteza.mahmoudisaber]

Actually in this case, segmentation fault is not due to memory error. BugWas uses GEMMA which accepts genotype and phenotype in BIMBAM format. For the phenotype it should be just one column without any headers, however, in the provided phenotype it is two columns with header. Delete the ID column and the header and it works fine.