Maximum number of contigs in a simulation

[teng-gao]

It seems mason_simulator crushes if there are >150 contigs in the fasta file.

mason_simulator -ir myfasta.fa -n 5000 -o left.fq -or right.fq

Error message:
MASON SIMULATOR

INITIALIZING__________________________________________________________

Opening reference and variants file ... OK
Initializing fragment-to-contig distribution ... OK
Initializing simulation threads ... OK
Opening output file left.fq ... OK
Opening output file right.fq ...Abort trap: 6

[marehr]

Hi @teng-gao,

can you attach your file myfasta.fa to the bug report? So that we can recreate your issue?

[Yixuan39]

Hello @marehr ,

I think I met a similar problem.

The command I used is:

mason_simulator -ir ./simGenome/Cattle/GCF_000003205.7_Btau_5.0.1_genomic.fna --out test/Cattle_sim_R1.fasta --out-right test/Cattle_sim_R2.fasta --out-alignment test/Cattle.sam --num-threads 16 -n 1000000 --illumina-read-length 150

The error I got is:

MASON SIMULATOR

INITIALIZING__________________________________________________________

Opening reference and variants file ... OK
Initializing fragment-to-contig distribution ... OK
Initializing simulation threads ... OK
Opening output file test/Cattle_sim_R1.fasta ... OK
Opening output file test/Cattle_sim_R2.fasta ...[1] 9565 abort mason_simulator -ir ./simGenome/Cattle/GCF_000003205.7_Btau_5.0.1_genomic.fna

The reference data I used is from:
https://www.ncbi.nlm.nih.gov/data-hub/genome/GCF_002263795.2/

Thank you!

[eseiler]

Mason seems to open a file handle for each fragment. The default limit is usually 1024 (ulimit -Sn).

If I increase my limit to 1048576 (ulimit -Sn 1048576, you can determine the absolute maximum via ulimit -Hn), simulation works.

Note: The file handle situation is a recurring problem with Mason. However, deferring the opening of the handles, i.e. only opening them when actually needed, would require a major restructuring of the code, which is not feasible for us. Hence, we focus on properly documenting the bahavior and adding checks with useful error output.

[Yixuan39]

Thank you for the answer! It works.