AI-DRIVEN ENZYME ENGINEERING
TA-Predict
Transaminase Prediction Server

Deep Learning-Based Transaminase Catalytic Performance and Stereoselectivity Prediction System

A fixed-deployment AI platform using deep feature fusion to predict apparent catalytic efficiency, high stereoselectivity and major product configuration of transaminase candidates.

Web: V1.0.0 Analysis pipeline: Standalone deployment Last updated: 2026-07-22
A. Submit a Single Prediction Job
Project information (optional)
Protein data (mandatory)
Reaction data (mandatory)
B. Batch Prediction
Upload candidate table
Download CSV Template
C. Model and Service Status

Verify that the fixed catalytic performance, high-stereoselectivity and configuration prediction modules are available before submitting company data.

Loading model status…
Complete health response 

        

      


      

        
D. Model Scope and Usage Notes

        

          

            
Catalytic performance

            

              The 3437-dimensional catalytic feature set is fused into a
              192-dimensional latent representation before family-aware
              regression. The server reports log10(kcat_app) and kcat_app.
            

          

          

            
High stereoselectivity

            

              The 2051-dimensional stereochemical feature set is fused into a
              304-dimensional latent representation. The fixed classifier
              reports the probability that ee reaches at least 99.8%.
            

          

          

            
Major configuration

            

              The fixed configuration module uses the same 304-dimensional
              fused representation and reports the predicted R/S class,
              together with P(R) and P(S).
            

          

          

            
Applicability

            

              Predictions are most reliable for enzyme families, substrates,
              mutation patterns and reaction conditions represented by the
              training distribution. Experimental verification remains required.