KILDA: identifying KIV-2 repeats from kmers

Abstract

High concentration of lipoprotein(a) [Lp(a)], a lipoprotein with proatherogenic properties, is an important risk factor for cardiovascular disease. This concentration is mostly genetically determined by a complex interplay between the number of kringle IV type 2 repeats and Lp(a)-affecting variants. Besides Lp(a) plasma concentration, there is an unmet need to identify individuals most at risk based on their LPA genotype. We developed KILDA (KIv2 Length Determined from a kmer Analysis), a Nextflow pipeline, to identify the number of kringle IV type 2 repeats and Lp(a)-affecting variants directly from kmers generated from FASTQ files. The pipeline was tested on the 1000 Genomes Project (n = 2459) and results were equivalent to DRAGEN-LPA (R ²= 0.92). In silico datasets proved the robustness of KILDA's predictions under different scenarios of sequencing coverage and quality. In brief, KILDA is a robust, open-source, and free-to-use pipeline that can identify the number of kringle IV type 2 repeats and Lp(a)-associated variants even when inputting low-coverage libraries.

Figure 1.

KILDA predictions against DRAGEN-LPA predictions for the number of KIV-2 copies on samples from the 1000 Genomes Project. For KILDA, 31-mers were used to make the predictions. The samples are colored by the presence of variants in samples as detected by KILDA.