SSPACE assembling Genome!

Question

[fufuyou]

Hi ,

I am assembling a genome using SSPACE. I met a problem as:

perl /scratch/snyder/f/fu115/Genome_assembly/SSPACE/41SSPACE-STANDARD-3.0_linux-x86_64/SSPACE-STANDARD-3.0_linux-x86_64/SSPACE_Standard_v3.0.pl -l libraries.txt -s contigs_abyss.fasta -k 5 -a 0.7 -x 0 -T 20 -b ecoli_scaffolds_no_extension
Your inserted inputs on [SSPACE_Standard_v3.0_linux] at Thu Nov 19 16:55:24 2015:
Required inputs: 
        -l = libraries.txt
                Number of paired files = 1
        -s = contigs_abyss.fasta
        -b = ecoli_scaffolds_no_extension

Optional inputs:
        -x = 0
        -z = 0
        -k = 5
        -g = 0
        -a = 0.7
        -n = 15
        -T = 20
        -p = 0

=>Thu Nov 19 16:55:24 2015: Reading, filtering and converting input sequences of library file initiated
------------------------------------------------------------
=>Thu Nov 19 16:55:50 2015: Storing contigs to format for scaffolding
LIBRARY lib1
------------------------------------------------------------
=>Thu Nov 19 16:55:51 2015: Reading contig file
=>Thu Nov 19 16:55:51 2015: Building BWA index for contigs
**************************************************
Process 'mapping reads' failed on Thu Nov 19 16:55:51 2015

I do not know what it is happen.

Thanks,
Fuyou

Answer 1

[sutturka]

I am sharing my experience with SSPACE as it may be useful for someone.

Data: We have multiple Paired-end and mate-pair libraries from a large eukaryotic genome. We performed assembly through ABySS and ran SSPACE on ABySS contigs to perform scaffolding and contig extension.

Issue: SSPACE scaffolding finished without any problems but contig extension (x = 1) was consistently failing.

Solution: Later, I found SSPACE uses unmapped reads from given library for extension and stores the reads in the memory. Therefore, the higher number of reads being stored in the memory was the issue for failing the SSPACE run. A simple solution is to divide your library into multiple files (I divided PE library file into 4 parts) and sequentially ran contig extension step to finish it successfully. Ultimately, a number of unmapped reads from the library (which are stored in the virtual memory) dictates the SSPACE run and run can be finished on low memory machines by splitting the files (in the case of MP library, I split my file into 14 parts). Hope this helps.

[AGE]

can you give an example of what your library.txt file looked like?

Answer 2

[tstrobaugh]

-T=20? Do you really have 20 cores or are you using hyperthreading? Back down to 8 or so. I'm only answering this question because I am having the same problem. Anyway, I believe this to be a "chmod" problem somewhere with BWA. I got it to work by replacing "BWA" with "bowtie" in the library file. Still trying to get bwa to work, problem is, it worked earlier and I didn't change anything.

[fufuyou]

It is the problem about RAM? I found if I change to -T=1, it worked well. But it is very slow. How about your work now?

Thanks,

Fuyou