Improving automated de-novo transcriptome definition in non-model organisms by integrating manually defined gene information

Abstract

Non-model organisms are of great ecological and economic significance, consequently the
understanding of their unique metabolic pathways by investigating their gene
expression profiles is essential. The bloom-forming alga Emiliania huxleyi
is a cosmopolitan unicellular photoautotroph that plays a prominent role in
the marine carbon and calcium cycle. Recently, genome sequences from several
key marine phytoplankton species have been sequenced and partially
assembled.  Nevertheless, there are many challenges in defining genes in
non-model organisms, where genomes are incomplete.  With the advent of next
generation sequencing technologies, cDNA short read sequences are generated
in ever increasing amounts, and tools for building transcripts abound. 
However, quality control of the transcript building process is rarely
performed if ever.  We used 63 manually defined genes, several
experimentally validated, in order to test the quality of the automated
transcriptome definition. We found that the automated pipelines missed genes
and artificially joined overlapping transcripts. In addition, E. huxleyi has
a very high percentage of non-canonical splice junctions, and relatively
high rates of intron readthrough, which caused unique issues with the
currently available tools and may indicate unique transcription machinery.
While individual tools missed transcripts, combining the results of several
tools improved the completeness and quality considerably.