FAQ

Is this an MPI program? Or do I need to set OMP_NUM_THREADS variable?

MEGAnE use MPI/OpenMP for threading through Python numpy module, however, it is controlled by MEGAnE. Therefore, you do NOT need to set up the 'OMP_NUM_THREADS' variable.

How many threads should I use?

In general, 2 to 4 threads is a good range for step 1 (discovery and genotyping of MEVs). While the program has steps with threading/multiprocessing, it also has steps limited to a single-thread. Therefore, the execution time does not decrease in a linear manner when you increase the number of cores used.

How much RAM does this require?

In step 0 (preparation of MEGAnE k-mer set), it requires 16 * (genome length) GB of RAM. In the case of human genome (3Gb), it requires ~50GB RAM.
In step 1 (discovery and genotyping of MEVs), it usually requires less than 15GB for 30x WGS, but it depends on the number of discordantly mapped reads.
In step 2 (joint calling), it requires 50GB RAM when merging ~10,000 subjects.

Are there any mount points in Singularity/docker container?

Yes, there are 10 available mount points: /mnt/mnt0, /mnt/mnt1, /mnt/mnt2, /mnt/mnt3, /mnt/mnt4, /mnt/mnt5, /mnt/mnt6, /mnt/mnt7, /mnt/mnt8, /mnt/mnt9.

E.g. in the case of Singularity:

export SINGULARITY_BINDPATH="/lustre/share:/mnt/mnt0,/lustre/private:/mnt/mnt1"

singularity exec MEGAnE.sif call_genotype_37 \
-i  /mnt/mnt1/path/to/input.cram \
-fa /mnt/mnt0/path/to/ref_genome.fa \
-mk /path/to/ref_genome.fa.mk \
-sample_name NA12878 \
-p 2