Scallop is an accurate reference-based transcript assembler. Scallop features its high accuracy in assembling multi-exon transcripts as well as lowly expressed transcripts. Scallop achieves this improvement through a novel algorithm that can be proved preserving all phasing paths from reads and paired-end reads, while also achieves both transcripts parsimony and coverage deviation minimization.
Scallop paper has been published at Nature Biotechnology. The datasets and scripts used in this paper to compare the performance of Scallop and other assemblers are available at scalloptest.
Please also checkout the podcast about Scallop (thanks Roman Cheplyaka for the interview). It is available at both the bioinformatics chat and iTunes.