Impact of sequencing data filtering on the quality of de novo transcriptome assembly

¹ Sevastopol State University, Sevastopol, Russia
² IBSS RAS, Sevastopol, Russia

Abstract

There are many assemblers with different algorithms that are used for de novo transcriptome assembly. At the same time, the filtering stage, which is one of the key stages, also has several approaches and algorithms. However, to date, there are only few studies on the effect of the degree of filtration on the de novo transcriptome assembly, specially for single-end reads. In this paper, we analyzed transcriptomes obtained using two of the most common software (rnaSPADES and Trinity), and also applied various approaches to the stage of filtering reads. The key differences between the two assemblies were shown and the parameters that were sensitive to the degree of filtering and the length of the input reads were identified. An efficient two-stage filtering algorithm was also proposed, which allows one to preserve the volume of input data as much as possible with the required quality of all reads after filtering and trimming.