Utility Of Abyss .Dot Files

Question

[Daniel Standage]

I just successfully completed a eukaryotic genome assembly with ABySS, and I can see it generated several .dot files, presumably for generating graphics with graphviz. My ${name}-scaffolds.dot file contains more than 10 million of lines, which as you might expect leads to some issues in practice. First, the bigmem machine I performed the assembly on doesn't even have enough memory to plot the graph with the dot program. Second, even if it did, it's highly unlikely that any useful visualization could be derived from the resulting graphic.

My question: what is the utility of the .dot files generated by ABySS? Are these primarily for working with small toy data sets, or is there some other utility that I've missed?

Thanks!

Answer 1

[Shaun Jackman]

Hi, Daniel.

The .dot files represent the sequence overlap graph of the contigs/scaffolds. The .dist.dot files represent the distances estimates between contigs/scaffolds based on paired-end information. These dot files are used to pass information from one stage of the assembly to the next. ABySS uses a fairly compact, memory-efficient format to store these graphs in memory. GraphViz uses a more versatile and general representation, but requires much more memory. As you noted, you can use GraphViz to visualize the graphs of small genomes. For larger genomes, you can pull out subsets of the graph for visualization (using grep for a start, or fancier tools potentially), or you can use graph visualization tools that are intended for large graphs, such as Gephi.

Here's a brand new wiki page to describe the various file formats of ABySS. It's terse for now, but it will be expanding. https://github.com/bcgsc/abyss/wiki/ABySS-File-Formats

Cheers, Shaun

Answer 2

[jts]

ABySS uses dot as its working file format - it will read and write dot files as it produces the assembly. These large dot files are not (primarily) for visualisation.