Tool for de novo assembly evaluation ??

Question

[Picasa]

Hello,

I've ran Soapdenovo, which produce a statistic file (scafStatistics).

However the N50 is different from the one produced by QUAST ...

Do you have an idea what is wrong ?

By the way, do you know a fast tool for computing assembler statistics such as N50, contig number etc.

Answer 1

[Chris Fields]

We've used the assemblathon script for this for quite a while, but there are likely better ones out there now.

http://korflab.ucdavis.edu/datasets/Assemblathon/Assemblathon2/Basic_metrics/assemblathon_stats.pl

Re: variation in N50, it may be dependent on whether the tool is reporting the stats on raw vs. a filtered (removing seqs less than a minimum cutoff). We generally perform both, and if at all possible use an estimated genome size so the metrics (NG50 in this case) can be compared more readily across assemblies.

Answer 2

[Picasa]

Is assemblathon script permors filtering before computing N50 ?

Cause I have different result compare to Soapdenovo and QUAST.

[Chris Fields]

For future reference, you should add this as a reply to my answer, not add this as another answer.

The script processes the sequences as is and performs no filtering; the other tools may give different results depending on whether they pre-filter or not. TBH I'm not sure.

EDIT: Also, make sure the N50 you're comparing is contig or scaffold-based (and whether there are any minimal N counts needed to split scaffolds into contigs).