Assemble

The next step in the workflow is to perform de novo assembly with MEGAHIT on the cleaned reads. Once the reads have been assembled into contigs, depth of coverage is calculated using Salmon. Captus makes it easy to process many samples in a consistent manner, automatically, and providing a comprehensive Assembly Control HTML report.

MEGAHIT is an extremely fast and robust de novo assembler that is very memory-efficient thanks to the use of a new data structure called succint De Bruijn Graph (sDBG). Most importantly, accuracy is not sacrificed because it was originally designed for assembling metagenomes. Since then, it has been shown to be an excellent generalist assembler as well.

If you are analyzing data with extremely high sequencing depth, or you are trying out parameters and want a quick result, Captus can subsample a fixed number of reads prior to assembly.

MEGAHIT only provides a quick estimation of contig depth of coverage, therefore we use Salmon to rapidly and accurately calculate the depth of coverage of each contig. Afterwards, contigs with little evidence (by default <1.5x) are automatically filtered.

If you assemble your reads in Captus you can also filter the contigs by GC content. For example, remove contigs with >60% GC will remove mostly bacterial contigs. The filtering by GC content and/or by depth of coverage can be repeated multiple times without needing to repeat the assembly and depth of coverage estimation steps.

Captus allows you the flexibility to also provide pre-assembled samples. However, we recommend that, whenever you have read data, to assemble it using Captus. For transcriptome assemblies for example, other assemblers will produce lots of redundant contigs (due to isoforms, alleles, etc.) which MEGAHIT tends to collapse into a single contig. This is ideal for phylogenomics (and perhaps also to build non-redundant reference transcriptomes).

Created by Edgardo M. Ortiz (06.08.2021)
Last modified by Edgardo M. Ortiz (18.12.2024)