CASE 1: AFFINITY OPTIMISATION
From thousands of candidates to actionable leads
Client: Twist Biosciences
Challenge
Twist needed to rapidly downselect from approximately 1,200 candidates to identify molecules suitable for downstream development.
A key consideration was balancing binding affinity with early indicators of developability risk, including aggregation and immunogenicity.
Co-pilot solution
Using EMLy Co-pilot, the team conducted affinity assessments, developability analysis, and humanness evaluation of all candidates through a single, guided computational workflow.
Outcome
-
Reduced 1,200 candidates to 45 high-potential leads
-
Completed in under 7 days
-
Avoided more than £50,000 in anticipated wet-lab costs
By focusing experimental resources on the most promising candidates, the client accelerated their optimisation timeline while reducing unnecessary spend and risk.
EMLy demonstrated >96% success in designing variants with improved yields and less aggregation compared to parent antibodies.
HAVE A SPECIALIZED PROJECT?
Connect with us to learn about custom solutions to meet your unique needs.
CASE 2: EXPRESSION OPTIMISATION
Resolving major manufacturability bottlenecks
Client: Vector Labs
Challenge
Manufacturability issues, such as poor expression yield and high aggregation propensity, are amongst the primary failure points for drug development.
Vector Labs was looking for a way to optimise these antibodies computationally before moving into costly lab testing.
Co-pilot solution
Vector Labs partnered with Etcembly to redesign and optimize several problematic antibody sequences using EMLy’s variant generation capabilities to improve manufacturability characteristics.
Outcome
-
>80% of redesigned variants showed improved expression and yield
-
Reduced aggregation risk across the majority of variants
-
Enabled faster progression to developable antibody leads
This collaboration highlights how integrated AI-driven design can accelerate early antibody optimisation, improving manufacturability before experimental evaluation.
80% of the EMLy-designed variants showed improved expression and fewer aggregation issues compared to the parental antibody without compromising the structure.
CASE 3: AI-GUIDED DESIGN
Redefining TCR engineering timelines
Client: Internal Etcembly program
Challenge
Engineering TCRs with sufficiently high affinity is critical for therapeutic efficacy, particularly for cancer targets that may be present at low abundance. Improving binding strength and stability has traditionally required lengthy, iterative experimental approaches, often taking years to achieve the affinity needed for reliable target engagement and downstream development.
The team wanted to demonstrate the potential of AI-driven design to accelerate this process and to discover high-quality therapeutic candidates.
Co-pilot solution
Etcembly scientists discovered and affinity engineered a TCR against the clinically relevant cancer antigen PRAME (Preferentially Expressed Antigen of Melanoma) using EMLy’s generative structural models to rapidly analyse and redesign candidate sequences in a fraction of the usual timeline.
Outcome
-
Affinity engineered a PRAME-targeting TCR with low-picomolar affinity
-
Reduced development time by ~50% compared to traditional approaches
-
Advanced ETC-101 into preclinical development
-
Improved expression to support manufacturability.
These results illustrate how AI-driven design can overcome long-standing bottlenecks in TCR engineering, speeding progression to preclinical development.
EMLy-predicted TCR structures (Model) and actual crystal structures (Crystal) showed a strong concordance
