Speeding up NUCmer for de novo contig to reference alignment

Question

[lkw222]

I have a de novo assembled mammalian genome (~300,000 contigs) that I want to align to a closely related existing reference genome using NUCmer from the MUMmer package. Subsequently, given the Coords output of the NUCmer alignment, I want to use something like OSLay to order and stitch together the contigs.

The problem is, NUCmer seems like it is going to take a very, very long time to run. I have split the reference genome up into single chromosomes and I am aligning all of my contigs to each chromosome (32 chromosomes, 32 jobs running at once). However, even the smallest chromosome has run over a day and is not complete. Any ideas on how to speed up the alignment?

I'm using:

nucmer --prefix bbu_vs_bt_ref_chr1 ./bt_ref_chr1.fa ./WB_2.0.fa

[apoorva]

Hello, Were you able to solve this problem? I am having the same trouble with my genome file. I have 16 chromosomes in the reference and 16 chromosomes in the query. I also have reduced the reference to 1 chromosome against 16 chromosomes in the query. But the time taken by the nucmer for each chromosome is like more than 1 day(still run has not completed). Any help would be appreciated.

Answer 1

[sst]

Never tried it, but http://bioinformatics.oxfordjournals.org/content/31/4/509.long claims to be an efficient drop-in replacement for MUMmer.

[cmdcolin]

just to improve google-ability of this page, the paper, in question here is "E-MEM: efficient computation of maximal exact matches for very large genomes "

Answer 2

[ALchEmiXt]

There is also a GPU optimized version you could try. How large is your reference and how large contigs? Monitor memory usage and increase memory... It needs to build a large suffix tree...