Can We Detect T Cell Receptors from Long-Read RNA-Seq Data?

T cells play an essential role in defense of the organism against pathogens and cancer. Efficient protection requires a vast repertoire of immune receptors, which is created by the V(D)J recombination process. There are multiple algorithms designed for the annotation of recombined T cell receptor (TR) sequences from traditional (short-read) RNA-Seq, however, none is adjusted for the long-read data. Here we intend to examine whether existing methods for TR sequences annotation using traditional RNA-Seq can be utilized for long-read sequencing data. ImReP, TRUST4, CATT and MiXCR algorithms were applied to data obtained by nanopore technology (PromethION). Adjustment of parameters was performed. The biggest number of CDR3 sequences was detected by the TRUST4 algorithm (20,599 unique TR sequences out of 73,904,478 total reads), representing 25% of the expected number of sequences. The distribution of annotated V and J genes was the same for MiXCR and TRUST4 algorithms and may be used to analyze the repertoire of V/J gene used in rearranged TR genes. Due to the high sequencing error rate of the analyzed sample (median read quality Q = 6.9), TR clonotype analysis is not suggested, and additional error correction steps are recommended for such analyses.

Automated summary

This study aimed to investigate the feasibility of utilizing existing methods for TR sequence annotation in long-read sequencing data. TRUST4 algorithm detected the largest number of CDR3 sequences, MiXCR and TRUST4 algorithms showed the same distribution in annotated V and J genes, which can be used to analyze the repertoire of V/J genes used in rearranged TR genes.