HGAP Parameters

[sagarutturkar]

Hi,

I am using HGAP through SMRTanalysis to assemble 2 MB microbial genome. I have 2 SMRT cells with more than 200x coverage. With HGAP2 from SMRTportal 2.1.1 i got around 20 contigs. Then I upgraded to v 2.2 and with HGAP 2 and 3 protocols I am always getting 24 contigs. I have also updated the target coverage to 15 since it is a microbial genome.

Also as per the pacbio recommendation "For samples with a lot of coverage (e.g. significantly greater than 100X coverage), you may see a larger number of contigs resulting from overwhelming the built-in contamination and chimera filtering that is part of the HGAP process. This can be addressed by using the ~100X longest subreads for HGAP, which can be selected by increasing the minimum subread length." I increased minimum subread length to 1000 and 2000. With subread 1000 I get 24 contigs assembly and with 2000 I get 180 contig assebly.


Based on this scenario I have some questions:

1. What parameter influence the HGAP other than target_coverage and min_subread_length?
2. I have updated min_subread_length to 1000/2000 in HGAP parameters. Is it right way to do it?
3. Why HGAP from 2.1.1. have better (20 contigs) results than 2.2?
4. I also have some Illumina data for this genome and with Spades I got 64 contigs. What other assemblers I could try for Only_pacBio or hybrid assembly?

Thanks
Sagar

[rhall]

What parameter influence the HGAP other than target_coverage and min_subread_length?

I find HGAP to be extremely robust, tweaking parameters is not going to significantly improve the assembly.

I have updated min_subread_length to 1000/2000 in HGAP parameters. Is it right way to do it?

Yes

Why HGAP from 2.1.1. have better (20 contigs) results than 2.2?

At this level the number of contigs is not a good measure of assembly quality. I would try to understand how the two assemblies compare.

I also have some Illumina data for this genome and with Spades I got 64 contigs. What other assemblers I could try for Only_pacBio or hybrid assembly?

I don't think it is worth trying other assemblers, time is probably better spent trying to understand the current assemblies. How big are the contigs? What do they blast against? Do they show any signs of overlap? Understanding why you are getting 20 contigs, i.e. is there a repeat that the size distribution of your library does not allow you to span through? Will help with the next step.

https://github.com/PacificBiosciences/Bioinformatics-Training/wiki/Evaluating-Assemblies

Richard.

[sagarutturkar]

Thanks

Thanks for the reply and suggestions. I will look into assembly details.