Re: [MUMmer-help] dnadiff very slow for Human

[Adam P. <aph...@gm...>]

Hi Davide,
dnadiff was primarily designed for microbial genome comparison and
currently does not scale well for large genomes. The 'delta-filter' step is
certainly one of the major bottlenecks. delta-filter scales by the number
of matches it has to analyze, so you can speed things along by reducing the
total number of matches. A few ways to do this:

1. Run nucmer in mum-reference mode to ignore repetitive alignment seeds
2. Increase the minimum match length and minimum cluster length (this will
reduce sensitivity to low-identity alignments)
3. Run delta-filter with the -l and -i options to filter alignments by
length and identity (these filters are quick, compared to -1/-m/-r/-q which
all require a dynamic programming step)

Once you have a filtered delta file using the above recommendations, you
can pass it directly to dnadiff using the -d option and it will skip the
alignment phase and process your delta filter directly--hopefully faster
than before.

Best,
-Adam




On Tue, May 20, 2014 at 1:34 AM, Davide VERZOTTO (GIS) <
ver...@gi...> wrote:

> Dear MUMmer users,
>
> We are trying to apply dnadiff for the analysis of breakpoints between our
> de novo Human genome assembly and the Reference genome, the latter divided
> into multiple chromosomes / separate files.
>
> We have already computed a NUCmer comparison between the two assemblies
> and the related delta file.  After this, we tried to compare all our
> scaffolds versus hg19 chromosome 1 using dnadiff, and the tool lasted more
> than 12 days (1 single core used, peak of 24 Gb RAM) before crashing (for
> internal server reasons), without writing any temporary file (apart from
> the log line "Filtering alignments") and presumably just trying to run
> "delta-filter -1".  Did you already face this problem?  Is there a way or
> script to speed up dnadiff for the Human genome comparison?
>
> Thanks and regards,
> Davide
>
> -------------------------------
> This e-mail and any attachments are only for the use of the intended
> recipient and may be confidential and/or privileged. If you are not the
> recipient, please delete it or notify the sender immediately. Please do not
> copy or use it for any purpose or disclose the contents to any other person
> as it may be an offence under the Official Secrets Act.
> -------------------------------
>
>
> ------------------------------------------------------------------------------
> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
> Instantly run your Selenium tests across 300+ browser/OS combos.
> Get unparalleled scalability from the best Selenium testing platform
> available
> Simple to use. Nothing to install. Get started now for free."
> http://p.sf.net/sfu/SauceLabs
> _______________________________________________
> MUMmer-help mailing list
> MUM...@li...
> https://lists.sourceforge.net/lists/listinfo/mummer-help
>