Any suggestion of running ABySS for about 50,000X coverage and paired-end 100 bp reads in genome size of 12.5 Mb - file size 770 Gb

[Giltae Song]

I'm about to run ABySS for assembly with data in 50,000X coverage and paired-end 100 bp reads. The genome is about 12.5 Mb. Do you have any suggestions to improve the assembly comparing to lower coverage data? Is it doable to run in regular paired-end mode of ABySS for the assembly for this data?  

[Ben Vandervalk]

Hi Giltae,

In response to your BioStar post, Brian Bushnell says:

"I do not think you will get a better assembly; more likely, a worse assembly.  Unless your coverage is very uneven, going over 100x or so typically starts to make the assembly worse, as there are an increasing number of exactly replicated sequencing errors, which create false branches in the deBruijn graph.  With thousands of X coverage, people typically normalize or subsample in order to achieve a better assembly.  Though it's possible that some metagenome, single-cell, or RNA-seq assemblers would be more tolerant of such high coverage."

In my opinion, this is good advice.

For what it's worth, ABySS automatically calculates its kmer coverage threshold for filtering out error kmers based on the kmer coverage histogram, so in principle your data set should assemble fine.  But you are probably not going to gain much from having all that extra coverage.

- Ben

On Mon, May 4, 2015 at 5:42 PM, Giltae Song <gilta...@gmail.com> wrote:

I'm about to run ABySS for assembly with data in 50,000X coverage and paired-end 100 bp reads. The genome is about 12.5 Mb. Do you have any suggestions to improve the assembly comparing to lower coverage data? Is it doable to run in regular paired-end mode of ABySS for the assembly for this data?  

--
You received this message because you are subscribed to the Google Groups "ABySS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to abyss-users...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.