Bridger: a new framework for de novo transcriptome assembly using RNA-seq data

Abstract

We present a new de novo transcriptome assembler, Bridger, which takes advantage of techniques employed in Cufflinks to overcome limitations of the existing de novo assemblers. When tested on dog, human, and mouse RNA-seq data, Bridger assembled more full-length reference transcripts while reporting considerably fewer candidate transcripts, hence greatly reducing false positive transcripts in comparison with the state-of-the-art assemblers. It runs substantially faster and requires much less memory space than most assemblers. More interestingly, Bridger reaches a comparable level of sensitivity and accuracy with Cufflinks. Bridger is available at https://sourceforge.net/projects/rnaseqassembly/files/?source=navbar.

Figure 1

Flowchart of Bridger. (a) The algorithm takes RNA-seq reads (single or paired) to assemble splicing graphs, each of which provides a complete representation of all alternative splicing transcripts for each locus. (b-d) Each splicing graph is processed independently. (b) Each edge in a splicing graph represents one splice junction. In this example, edges 1 and 3 are compatible, while edges 3 and 4 are not compatible. (c) A compatibility graph. (d) A minimum path cover model is applied to recover a minimal set of transcripts that could be tiled together through overlapping sequence reads and ‘explain’ all observed junctions in a splicing graph.

Figure 2

Splicing graph construction. (a) Splicing graph after branch extension. The red k-mer (k = 5) ATCAG on the left is a bifurcation 5-mer because there is an unused 5-mer TCAGC in the hash table that provides an alternative extension. Extend this 5-mer to a new contig until it cannot be further extended. We check the last 4-mer of this branch to see if there is a matching 4-mer in the current splicing graph. If so, another bifurcation 5-mer is found (for example, the red 5-mer CTAGC). (b) A modified splicing graph by merging the k-1 overlapping nucleotides (4-mer CTAG) and adding a new directed edge between two bifurcation k-mers.

Figure 3

Comparison of the number of full-length reconstructed reference transcripts for (a) dog, (b) human, and (c) mouse.

Figure 4

Comparison of accuracy for (a) dog, (b) human, and (c) mouse.

Figure 5

Run time and RAM usage for each assembler in (a) dog, (b) human, and (c) mouse. Same parameter values are used for all assemblers: k = 25 and CPU = 6.

Figure 6

CPU time for each assembler in (a) dog, (b) human, and (c) mouse.

Figure 7

A novel gene containing 10 exons was assembled by all assemblers. Interestingly, all de novo assemblers captured longer UTR than the reference-based assembler Cufflinks.