Phylogenomics from Whole Genome Sequences Using aTRAM

Abstract

Novel sequencing technologies are rapidly expanding the size of data sets that can be applied to phylogenetic studies. Currently the most commonly used phylogenomic approaches involve some form of genome reduction. While these approaches make assembling phylogenomic data sets more economical for organisms with large genomes, they reduce the genomic coverage and thereby the long-term utility of the data. Currently, for organisms with moderate to small genomes ($<$1000 Mbp) it is feasible to sequence the entire genome at modest coverage ($10-30\times$). Computational challenges for handling these large data sets can be alleviated by assembling targeted reads, rather than assembling the entire genome, to produce a phylogenomic data matrix. Here we demonstrate the use of automated Target Restricted Assembly Method (aTRAM) to assemble 1107 single-copy ortholog genes from whole genome sequencing of sucking lice (Anoplura) and out-groups. We developed a pipeline to extract exon sequences from the aTRAM assemblies by annotating them with respect to the original target protein. We aligned these protein sequences with the inferred amino acids and then performed phylogenetic analyses on both the concatenated matrix of genes and on each gene separately in a coalescent analysis. Finally, we tested the limits of successful assembly in aTRAM by assembling 100 genes from close- to distantly related taxa at high to low levels of coverage.Both the concatenated analysis and the coalescent-based analysis produced the same tree topology, which was consistent with previously published results and resolved weakly supported nodes. These results demonstrate that this approach is successful at developing phylogenomic data sets from raw genome sequencing reads. Further, we found that with coverages above $5-10\times$, aTRAM was successful at assembling 80-90% of the contigs for both close and distantly related taxa. As sequencing costs continue to decline, we expect full genome sequencing will become more feasible for a wider array of organisms, and aTRAM will enable mining of these genomic data sets for an extensive variety of applications, including phylogenomics. [aTRAM; gene assembly; genome sequencing; phylogenomics.].

A Guide to Carrying Out a Phylogenomic Target Sequence Capture Project

“…Pipelines also differ in terms of which part of the postsequencing workflow they cover. Some pipelines are particularly focused on the specific steps of assembly and recovery of longer compound sequences from the read data (sequence engineering), such as aTRAM (Allen et al, 2017) and HYBPIPER (Johnson et al, 2016). Other pipelines are more focused on guiding users through the complete process from cleaning raw sequencing reads to producing data structures that can be readily used for phylogenetic inference (e.g.…”

The role of parasite dispersal in shaping a host–parasite system at multiple evolutionary scales

“…Both types of lice also form monophyletic groups within their respective genera (Johnson et al, ), which makes interpretation of evolutionary history straightforward. Obtaining genomic‐level data is very feasible for these lice, as recently published genomic studies on avian lice have established pipelines for assembling data appropriate for both phylogenetic and population genetic analysis (Allen et al, ; Boyd et al, ; Sweet et al, ).…”

Genomic sequence capture of haemosporidian parasites: Methods and prospects for enhanced study of host–parasite evolution

“…We next used custom scripts written in Perl and Python to keep only the contigs from the last atram iteration (available at https://github.com/juliema/phylogenomics_scripts; “getlastiteration.pl”), compare and align them to the translated exon sequences for the reference H. tartakovskyi using exonerate 2.2.0 (Slater & Birney, ) and stitch together any exons that were broken into multiple contigs (as described in Allen et al, ; https://github.com/juliema/exon_stitching). Because atram performs iterative assemblies that can extend outward from the original reference sequence, the last iteration is most likely to have the most complete, longest contigs.…”