Visualizing contig alignment

Question

[arunprasanna83]

I have two genome assemblies created using different platforms for same species. I am doing some post analysis to check the performance of assemblers. For this, I have the following:

1. Contig files (multi-fasta) for reference (910 MB) and query (510 MB)
2. Blast hits (tabular format) for reference vs query.

Here, I would like to make a circular track for Reference genome and see how the hits of query assembly distribute. Please suggest the possible methods to do it. I looked for UCSC genome browser, where it displays alignment for pre-defined set of organism. My organism is not in their list.

Kindly help.

Thanks in Advance.

[lieven.sterck]

Circos might be an option or D-genies

[arunprasanna83]

D-genies crashes while I upload the data in .gz format.

[lieven.sterck]

that's a pitty.

I'm not on the development team myself ;-) but I know some of them, I'll pass on the message. But don't let that hold you back to get in touch with them (== send bug report) yourself.

[arunprasanna83]

Thanks @lieven.sterck. I did send them an email 3 weeks back. But did not get any response !. So moved on to explore other options.

[arunprasanna83]

There is a bigger problem here. Most of these programs crash due to large file size. i.e File 1 has 44k contigs (510MB) and File 2 has 5900 contigs (910 MB). To make it easier, I tried to merged respective files into single long sequence. But still, the programs crashes ! Any solutions to handle these big data ?

[GenoMax]

Are you sure the problem is not the hardware you have access to?

[arunprasanna83]

Sure ! I am doing it in my linux workstation with 150 GB RAM and 40 Cores.

Answer 1

[Philipp Bayer]

Are dotplots not an option?

Minidot is very fast, but in my experience doesn't work that great with 'bad' assemblies: https://github.com/thackl/minidot

Symap is a bit older and slower, but works better with low quality assemblies (again, in my experience), and can also make a circular plot http://www.agcol.arizona.edu/software/symap/

Circos should be able to take both output files with some fiddling. You should be able to take minidot's or symap's alignment output files, write a tiny parser to turn them into the tabular format circos wants, and then run bundlelinks on that.

[arunprasanna83]

Thanks ! Out of these, symap is doing a neat job !.

[arunprasanna83]

Symap worked for comparing small vs big assembly. But it is still running for other case 'big vs big' assembly (910 Mb, 5931 contigs) for self comparison over 5 days !. Any tips ?

[Philipp Bayer]

I usually remove small contigs (<10kb) because these will only add chaos to the graph anyway.

Answer 2

[harish]

If you have mummer alignments, you can probably use Circlize or AliTV.

Alternatively, you can also use Gepard or minidot etc to view dot-plot alignments.

[arunprasanna83]

Will try Circlize or AliTv. Because, I tried both Gepard & minidot. Both failed due to large file size !

[arunprasanna83]

As I mentioned in comments above, AliTV is also working fine for smaller files. But for bigger files it is taking forever. I guess there is no parallelization option.