PhyMMR: Phylogenetics through markov mapped reads - Revolutionizing dataset preparation for phylogenetic analyses

Investigation into phylogenetic DNA dataset preparation pipelines revealed unrecovered data, improper contig mapping, and transcript re-use. PhyMMR, Phylogenetics through Markov Mapped Reads, was developed to resolve these issues.

PhyMMR allows for the combination of transcriptome, genome, AHE and UCE data. Hidden Markov Modeling using HMMER, in combination with BLAST, is used to map reads (or assemblies) to reference orthologs. These mappings undergo a internalized cleaning process involving alignment, end-trimming, outlier removal, and a cross-contamination check prior to merge. This pipeline brings significant improvements in data recovery, accuracy, ease of use, and lowers time required to produce an analysis ready dataset from months to days.

Using published UCE, AHE and Transcriptome data, we demonstrate PhyMMR increases target data recovery by 1.5x to >20x compared to original study results. Additionally, we show that mapping reads directly against reference orthologs is superior to mapping assembler output. Use of PhyMMR resolves poorly supported nodes previously thought intractable. To illustrate this strength, we present a phylogeny of Acalyptratae constructed using publicly available data.