Combined computational triage and Pichia pastoris expression workflow for protein variant screening

Overview

This proposed workflow addresses a common bottleneck in protein engineering and biopharmaceutical development: the high cost and manual effort of expressing and testing hundreds or thousands of protein variants. It combines two complementary stages into an iterative pipeline. First, computational methods triage the variant list by estimating stability, aggregation propensity, and solubility, so that likely poor performers are removed before any laboratory work. Second, the prioritised candidates are expressed in the eukaryotic host Pichia pastoris and characterised with standardised assays for expression level, soluble yield, aggregation, stability, and purity. Measured results are then fed back into the computational models, improving their accuracy with each cycle.

The value proposition is a reduction in hands-on effort and cost per candidate, combined with the ability to scale screening to very large variant panels. The workflow is designed as a filter rather than a replacement for experiment: computational predictions identify promising candidates, while laboratory expression and characterisation provide the definitive data. This makes it relevant to any research or development programme that needs to screen many protein variants for stability, solubility, and expression.

Technical specifications

Stage 1: Computational triage

• Structure-based modelling estimates stability and aggregation propensity.
• Molecular dynamics simulations provide additional ranking information on protein behaviour.
• Weak candidates are removed before any wet-lab work, reducing the number of variants taken forward.
• The pipeline can score thousands of sequences at low marginal computational cost.

Stage 2: Pichia pastoris expression and characterisation

• Prioritised candidates are expressed in Pichia pastoris, a scalable eukaryotic host.
• The host's natural protein secretion simplifies recovery and purity assessment.
• Standardised assays measure expression, soluble yield, aggregation, stability, and purity.
• Results are compared with computational predictions and used to refine the models in an iterative feedback loop.

The workflow is intended to be benchmarked against existing workflows in a pilot phase, measuring metrics such as candidates processed per week, hands-on hours per candidate, and hit rate.

The proposing team brings 12+ years of structure-based design, a PhD in molecular simulation, and hands-on Pichia pastoris expression experience.

Technology readiness level

The workflow is currently at an early research stage. The underlying computational methods and Pichia expression techniques are established individually, but the integrated pipeline has not yet been built or validated as a complete workflow. A six-month pilot is proposed in four phases: agree on a representative construct set and success metrics; build the computational triage pipeline and score the panel; express and characterise prioritised candidates; and compare predictions with measurements to refine the models. Go/no-go checkpoints would be used after the computational and expression phases, and results would guide any scale-up.