Benchmarking immunoinformatic repertoire assemblies from bulk RNA-seq and PCR-based V(D)J sequencing using PCR-free SMRT RNA sequencing.

Immune repertoire sequencing (RepSeq) is a crucial immunomonitoring tool, requiring high sensitivity, accuracy, and minimal bias. Although V(D)J mRNA sequencing using 5' rapid amplification of cDNA ends (RACE-RepSeq) is broadly used, its reliance on PCR may create distortions. Alternatively, in silico repertoire-reconstruction (ISRR) tools, computing bulk RNA-seq data, raise concerns regarding accuracy. We compared RACE-RepSeq and ISRR with an advanced PCR-independent approach based on single-molecule real-time (SMRT) sequencing of full-length transcripts. All three approaches applied to immunoglobulin transcripts yielded consistent repertoire-level metrics, including V/J gene usage, CDR3 features, and diversity indices. However, discrepancies emerged in somatic hypermutation assessment, isotype distribution, and at the individual clone level, including incomplete clone overlap and inconsistent isotype assignment, even among abundant clones. These methodological biases exceeded sampling effects. Overall, our results delineate the limitations of each approach in samples from B cell-rich tissues and support SMRT sequencing as a useful reference for immune repertoire benchmarking.