Team Training and Workshops

The most effective protein engineering teams are the ones where bench scientists understand computational tools well enough to use them in their daily work, and computational scientists understand the wet lab well enough to design useful experiments. But building that cross-functional fluency is hard. Most bioinformatics courses are too theoretical for bench scientists who need practical answers, and most bench training ignores the computational tools that have become essential to modern protein engineering.

The workshops I run are designed specifically for this gap. They are hands-on, project-based sessions where participants work with real data from their own programs. A typical workshop covers structure prediction with AlphaFold or Boltz-2, binding site analysis, sequence conservation mapping, and how to interpret computational scores in the context of experimental data. Participants leave with a working understanding of what these tools can tell them, what they can't, and how to integrate computational results into their experimental decision-making.

Training modules are customized to your team's needs, but common topics include: structural biology fundamentals for non-crystallographers (reading PDB files, understanding B-factors, interpreting electron density); protein structure prediction (when to trust AlphaFold, how to read confidence scores, what pLDDT and PAE actually mean); sequence analysis (building MSAs, reading conservation plots, identifying functional hotspots); and computational protein design (what RFDiffusion and ProteinMPNN do, how to evaluate designed sequences, what makes a good design objective).

For teams that are further along, I offer advanced sessions on machine learning for protein engineering: building custom predictors from experimental data, using protein language model embeddings for variant scoring, and designing ML-guided directed evolution campaigns. These sessions assume basic Python familiarity and are aimed at scientists who want to build and deploy their own models rather than relying on external tools or consultants.

Each workshop includes a practical exercise using the team's own targets or datasets. This is not an academic seminar with toy examples—it is a working session where participants generate outputs they can use in their ongoing programs. The format is typically a half-day or full-day session, delivered on-site or remotely, with follow-up materials and reference documentation.

The goal of training is not to create a permanent dependency on external consultants. It is to build enough internal capability that your team can run computational workflows independently for routine tasks, and engage consultants only for genuinely novel or complex problems. A bench scientist who can run an AlphaFold prediction and interpret the result doesn't need to wait three weeks for a computational collaborator to do it for them. A project leader who understands what binding affinity predictions can and can't tell you makes better decisions about which candidates to advance.

The most successful training engagements are followed by a short advisory period—typically one to two months—where participants can ask questions as they apply what they learned to their own projects. This bridges the gap between understanding a tool in a workshop setting and using it confidently in production, and it ensures that the investment in training translates into lasting capability rather than a one-time event.