Multi-K-Mer de novo Transcriptome Assembly, Validation, and Count Summarizing for Four Plant Taxa

Large genomes, polyploidy, and repetitive DNA are common obstacles in the assembly of plant genomes making de novo transcriptomes valuable genomic resources. To generate high quality de novo transcriptomes, we developed a custom assembly workflow for illumina and 454 RNA-Seq reads. The primary components of the workflow include stringent pre-cleaning; Oases multi-k-mer assembly for Illumina reads; MIRA assembly for 454 reads; and MIRA to cluster resulting contigs. We tested the workflow using data from four transcriptomes, two polyploid monocots and two dicots. Assemblies were validated using number of contigs, cumulative length of contigs, and N50 metrics. Ortholog hit ratios (OHR=length of alignment:length of proteins from a close relative) were calculated to estimate assembly fragmentation. Each clustered assembly had a high N50 (2.6-3.2 kb) and a high percentage of hits with an OHR >= 0.8 (52-65%) suggesting the workflow produced high quality assemblies.

For de novo transcriptomes, there are few standalone programs that summarize aligned read counts for input into EdgeR or DeSeq. Although popular, HTSeq allows partial length single-end alignments but does not allow alignment of one out of two mates. We developed the custom script Count_reads_denovo.pl, for de novo RNA-Seq projects. Count_reads_denovo.pl uses a model similar to featureCounts, a reference-based summarizer, to leverage paired-end data even where only one mate aligns. The script filters read counts by mapping quality (MAPQ). Scripts used in the above workflow, as well as Count_reads_denovo.pl are available at https://github.com/i5K-KINBRE-script-share/.