Cross-platform Assembly (PacBio & Illumina)

[bonifera]

How many of you have successfully merged and assembled a genome using PacBio reads merged with Illumina reads?

I'm wondering the best way to approach this.

Any help you can give would be greatly appreciated.

[sisch]

In very general:
I remember someone suggested curing the long PacBio reads with the accurate Illumina reads by mapping. And then assembling the resulting high quality consensus reads in a traditional way (i.e. assembler of your choice). In essence, you could call that a reference based pre-assembly where the reference are the PacBio reads.

Unfortunately I can't give any reference, because I can't remember where I heard this.

Best,
Simon

[Scaon]

Reference

http://www.nature.com/nbt/journal/v30/n7/full/nbt.2280.html

[bonifera]

@sisch Thanks for your input though! It certainly gives us a starting point.

@Scaon Awesome! Thank you so much!

[lhon]

Hgap

See also HGAP, the nonhybrid PacBio-only assembly approach:

http://www.nature.com/nmeth/journal/vaop/ncurrent/abs/nmeth.2474.html

[phenotype]

Why bother

Hi Simon,

You must be talking about the PacBioToCA script (http://sourceforge.netapps/mediawiki...tle=PacBioToCA) where you have to convert fastq files to .frg files and use those as input to Celera Assembler.

But I agree with lhon. Why bother when there is a much simpler workflow using long reads only with HGAP in SMRT Analysis 2.0.0? (http://pacbiodevnet.com/)

[krobison]

What scale genome are you sequencing?

I've found the HGAP pipeline very useful, but you do need to get high coverage (100X+), which may not be economical on larger genomes. Illumina may also be valuable for correcting indels that Quiver doesn't clean up. I've had less luck with PacBioToCA, in part because some parts of the genomes of interest simply don't show up in the Illumina data.

If HGAP isn't quite getting the genome closed due to coverage limitations, I have had luck closing some gaps by using Minimus2 to merge an Illumina assembly and a PacBio one. Directly assembling the two sets with MIRA or CA is on my to-do list.

[bonifera]

@phenotype

We're trying to correct the PacBio long read and high error rate with the illumina short read and low error rate. These are the data sets that we have