A comprehensive assessment of tandem repeat genotyping methods for Nanopore long-read genomes

Benchmark of 7 tandem repeat genotypers on Nanopore long-read genomes, using four validation strategies

The authors reviewed 25 tandem repeat genotyping tools and selected seven actively maintained methods for benchmarking on publicly available Oxford Nanopore genome sequencing data from more than 100 individuals. Performance was assessed across 43,009 genome-wide loci using four complementary strategies: concordance with haplotype-resolved Human Pangenome Reference Consortium assemblies, Mendelian consistency, cross-tool consistency, and sensitivity to molecularly confirmed pathogenic expansions. Most methods showed high concordance with assemblies, with better accuracy using R10 than R9 chemistry; accuracy declined with allele length, and most tools performed worse on homopolymers, heterozygous loci and alleles differing from the reference. Assembly concordance and Mendelian consistency did not predict sensitivity to pathogenic expansions, and no single tool performed best across all assessments.

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

The useful message for the laboratory is methodological: length accuracy overestimates genotyper performance, and a tool that excels against assemblies can miss a pathogenic expansion — the latter is the criterion on which to judge a tool intended for diagnosis. The abstract names neither the top-ranked tools nor the number of pathogenic expansions tested, so the full article is needed before choosing. The conclusions apply to Nanopore and do not automatically transfer to other long-read technologies.

Analysis by Dr Thibaut Benquey