Estimation of Variation within Transcriptome de novo Assemblies, co-assemblies of Paralogs, Heterozygosity or Polyploidy Possibly Accumulated Within the Isotigs of a Newbler Assembly

Transcriptome references built from sequencing reads provide foundations for gene discovery and functional genomics studies, sequence-based marker development and quantitative expression analysis, especially in species and accessions lacking the benefit of a whole genome reference.  Use of a common set of assembly parameters cannot adequately account for the breadth of natural diversity among transcriptome profiles, and yet most commonly, consensus transcript sequences are not further scrutinized.  Most notably with sequence data from heterozygotes, polyploids, and taxa with large gene families, transcript assemblers sometimes compress sets of paralogs, homoeologs and/or alleles into a single isotig, or transcript, leading to an incorrect picture of the gene expression profile.

 Here we describe an algorithm that estimates variation lost in an isotig consensus sequence, defined as error within the isotigs during assembly. Single nucleotide polymorphisms are called after mapping back to each isotig its individual reads and applying filters based on sequence quality. Based on the possible combinations of overlapping reads, full-length variants that could represent different transcripts are determined and scored based on the number of reads supporting the variant.  

We demonstrate the utility of our approach in its application to Newbler assemblies of allohexaploid, as well as homozygous and heterozygous diploid, plant transcriptomes.  By curating and scoring all elements in a transcriptome assembly, our approach enables the researcher to detect intra-isotig variation with biological relevance that may have otherwise been missed.  A portion of the data that requires greater attention then can be selectively reassembled to achieve greater separation of distinct transcripts.