Webinar: Can you actually trust computational TCR-Epitope annotations?

Computational annotation of T-cell receptors are a promising replacement of wet-lab assays to determine what T-cell receptors effectively recognise. But how reliable are these tools, and which approach should you trust?

In this webinar, we walk through the three major strategies for TCR-antigen annotation and the evidence behind each.

Strategy 1: Database matching - and why it's not enoughThe most intuitive approach is to look up your TCRs in a curated database. We'll show why this strategy breaks down in practice and might not be as trustable as intuitively expected.

Strategy 2: Seen epitope algorithms - the current standardDedicated ML algorithms were built to solve database matching problems. We'll cover how they work, what the independent IMMREP23 benchmark revealed about algorithm performance, and the multi-layered validation evidence — from external experiments to patient-level diagnostics — that establishes when you can confidently trust a prediction.

Strategy 3: Unseen epitope models - the next frontierPredicting binding to epitopes never seen in any training set remains the field's biggest unsolved challenge. We'll map where the frontier stands after IMMREP25, why it's so hard, and the hybrid modelling and data generation strategies that are trying to close the gap.

This webinar is designed for immunologists, bioinformaticians, and biopharma teams evaluating computational TCR annotation for their pipelines.