kTypeR – accurate and efficient alignment-free HLA genotyping using nanopore sequencing

Attempts to leverage nanopore sequence data for HLA genotyping are rapidly gaining traction in the HLA community as the technology promises easy and cost-effective library preparation and reads that cover the full extent of HLA Class I and Class II genes. Current typing strategies using nanopore sequences all rely on classical alignment-based methods. Here, we present ktypeR, which uses a fundamentally different strategy for genotyping. KtypeR compares all k-mers of each individual read to the pre-processed IPD-IMGT/HLA database and infers the most likely true alleles by majority vote.

kTypeR was developed using Minion, R10 pore with a total of 384 samples, each containing the six major HLA loci. A maximum of 243 samples are multiplexed using one barcode per sample. kTypeR is multi-threaded and can process 10,000 reads in 5 minutes on 8 cores

kTypeR achieved an outstanding accuracy of > 98 % alleles correctly genotyped in the 4th field using the R10 pore. As a conclusion, kTypeR shows great potential to outperform existing classical tools in speed and accuracy and demonstrated the feasibility of using nanopore data for mid- and high-throughput HLA genotyping at full resolution.