Determining SPADEs Cx with multiple nodes

[kingcohn1]

I am hoping to determine the depth of coverage at each contig, but I'm hung up by the headers. Each contiguous sequence has a top line of:

>NODE_1_length_56345_cov_1.86348

Now, I know how to get coverage for this region using: Cx=Ck*L/(L-k+1)

But, I'm confused as to why there are multiple NODE_1's. I hope to get total coverage for each contig, so how can I distinguish between multiple NODE_#'s?

[Brian Bushnell]

There should only be a single node 1; if you see multiple, please report that to the devs. I don't personally recall ever seeing that happen.

Also, you'll usually get a more accurate idea of the coverage by mapping the input reads to the assembly than from Spades' coverage value. For one thing, Spades is a multi-kmer assembler, so it's not even clear what k means in this context. But for another, well... mapping to calculate coverage is simply more accurate even for single-kmer assemblers.

[kingcohn1]

Hi Brian and thanks for responding to me, again! I think my egrep regex was appending all of the node_1's from all of my draft assemblies...So, what should I use to map my reads to the contigs-LASTz? The next step in my pipeline is to create larger scaffolds via a reference guided algorithm (ragout or pyscaf, do you have any recs?) and then compare outputs in mauve.

PS thanks for bbmaps, I recommend it when I can.

[GenoMax]

Since you are familiar with bbmap use it to map the original reads back to the contigs.

[Brian Bushnell]

To capture the per-contig coverage using BBMap, just add the flag "covstats=", e.g.:

bbmap.sh in=reads.fq ref=contigs.fa covstats=covstats.txt