This is a test version of Biostars. For the public version, visit https://www.biostars.org.
BWA. Allocate memory

Dear all,

I have changed the work station and now I am using Centos 7 as operating system.

And if I write this I found that I think I do not have such memory as run a pipeline (in python) to analyse an exome-seq

> free -h
              total        used        free      shared  buff/cache   available
Mem:           7,5G        1,9G        665M        435M        5,0G        4,9G
Swap:          1,9G        1,1G        816M

I've tried with a little sample but when I start mapping the fastq files (using BWA), I reach that:

---------------------------------------------------------------------------------------
                               Mapping fastq files (BWA)                                      
---------------------------------------------------------------------------------------
[M::bwa_idx_load_from_disk] read 0 ALT contigs
[bwa_idx2mem] Failed to allocate 4705743064 bytes at bwa.c line 346: Cannot allocate memory

So my question is how much memory do you recommend me to have, or if there is another way.

thanks!

P.s. I am talking about human samples, using hg19 as genome ref.

next-gen sequencing alignment bwa memory
• 5,993 views • link
written by windsur • windsur

You are really close to the memory limit with your 5G available to map reads against human genome using BWA. What are your command lines to index your reference genome and to align your reads ?

You can take a look to HISAT which has AFAIK the lowest memory requirement.

• 1 views • link
written by Bastien Hervé Bastien Hervé

Thanks! I will try to add more memory and also what you said.

what I use is:

call('bwa mem -t' + str(args.threads) + ' -R "@RG\tID:' + sample_name + '\tLB:library\tPL:illumina\tPU:library\tSM:' + sample_name + '" ' + genome_ref + ' ' + forward_paths[i] + ' ' + reverse_paths[i] + ' > ' + sample_path + '/' + sample_name + '_bwa.sam',shell = True)

after that I take the sam file and using samtools I create the sorted bam file:

call("find " + sample_path +  "*.sam | parallel --no-notice -j" + str(args.parallelization) + " 'samtools sort {} -O BAM -@ " + str(args.threads / 2) + " -o {}_sorted.bam && samtools index {}_sorted.bam'", shell = True)
• 1 views • link
written by windsur • windsur

Did you successfully index your genome_ref ?

In your python code try to print your call command and copy/paste them here, it's really hard to investigate with all these variables

• 1 views • link
written by Bastien Hervé Bastien Hervé

For exome sequencing, better stay with BWA mem, as it is the standard aligner for many downstream tools, including most variant and SV callers.

• 1 views • link
written by ATpoint ATpoint

True, that was just a memory test first, and if OP has no other choice HISAT would have be one solution

• 1 views • link
written by Bastien Hervé Bastien Hervé

Thanks Bastian, I will try with HISAT too. And to index my reference I have followed the steps of GATK

• 1 views • link
written by windsur • windsur

The question about your index was, was it sucessfull ?

• 1 views • link
written by Bastien Hervé Bastien Hervé

hey Bastian, yes it was sucessfull

• 1 views • link
written by windsur • windsur

Exactly! after using samtools I unload the genome reference:

call('bwa shm -d',shell = True)

Or should I not use samtools for the analysis you mean?

• 1 views • link
written by windsur • windsur

0 answers

No answers yet.

Log in to answer this question.

More posts like this