mRNAmarkup: quality control and annotation of de novo transcriptome assemblies

<p><em>De novo</em> assembly of RNA-Seq data produces pseudomolecules representing the best estimate of the actual ensemble of transcript isoforms in a given RNA sample. The assembly process typically introduces errors and artifacts, including fusion of overlapping transcripts into chimeric pseudomolecules. In the absence of cognate genomic sequence data, resolution of such errors will remain tentative. However, comparative genomics approaches can provide reliable suggestions to correct an initial assembly. We have developed mRNAmarkup as a lightweight mRNA annotation tool that 1) identifies potential contaminants in the RNA sample, such as vector or incidental bacterial or microsporidial sequences; 2) identifies and resolves likely chimeric assemblies; 3) indicates pseudotranscripts representing probable full-length mRNAs; 4) displays sets of alternatively spliced transcripts; and 5) annotates transcripts based on significant similarity to entries in reference protein sets. The final product of mRNAmarkup is a set of vetted and annotated transcripts. Advantages of mRNAmarkup relative to similar software are its enhanced vetting capabilities and its lightweight implementation, including parallelization of all compute-intensive steps, as well as portability, modularity, and extensibility. Use of the program will be discussed by re-analysis of several recently published RNA-Seq assemblies.