Alignment-free filtering for cfNA fusion fragments

Abstract

Motivation: Cell-free nucleic acid (cfNA) sequencing data require improvements to existing fusion detection methods along multiple axes: high depth of sequencing, low allele fractions, short fragment lengths and specialized barcodes, such as unique molecular identifiers.

Results: AF4 was developed to address these challenges. It uses a novel alignment-free kmer-based method to detect candidate fusion fragments with high sensitivity and orders of magnitude faster than existing tools. Candidate fragments are then filtered using a max-cover criterion that significantly reduces spurious matches while retaining authentic fusion fragments. This efficient first stage reduces the data sufficiently that commonly used criteria can process the remaining information, or sophisticated filtering policies that may not scale to the raw reads can be used. AF4 provides both targeted and de novo fusion detection modes. We demonstrate both modes in benchmark simulated and real RNA-seq data as well as clinical and cell-line cfNA data.

Availability and implementation: AF4 is open sourced, licensed under Apache License 2.0, and is available at: https://github.com/grailbio/bio/tree/master/fusion.

Fig. 1.

An overview of AF4 workflow. Detailed descriptions are given in the main text

Fig. 2.

Fusion event involving gene pairs (g₁, g₂) and readpairs (r₁, r₂). The green line denotes the fusion transcript derived from genes g₁ and g₂ with fusion junction point denoted by x_b. (a) Neither r₁ nor r₂ spans x_b, (b) r₁ but not r₁ spans x_b and (c) both r₁ and r₂ span x_b

Fig. 3.

Fragment generation by stitching and overhang trimming of readpair (r₁, r₂). (a) r₁ and r₂ are represented as arrows facing each other denoting the forward and reverse complement strands. The green bars denote one of the shared kmers between them, which is an anchor for suffix—prefix alignment. The stitched fragment is a concatenation of prefix of r₁, overlap and suffix of r₂. (b) When r₁ and/or r₂ extends beyond the $5\prime$ region of the other read, the overhang is trimmed, and f is the overlap. (c) When r₁ and r₂ cannot be merged, f is a concatenation of r₁ and reverse complement of r₂

Fig. 4.

Computing maximum coverage of fragment f for a gene pair (g₁, g₂). g₁ and g₂ are two genes inferred to cover regions of f. g₁ covers regions $s_1,s_2,s_3$, and g₂ cover regions $s_4,s_5,s_6$. $[x_i,x{\prime }_i)$ are start and end positions of f for region s_i