Cofactory

Cofactory predicts enzyme cofactor specificity from primary amino acid sequences to support selection of enzymes and optimization of cofactor balance in metabolic engineering.

Key Features:

• Sequence-based prediction: Predicts cofactor specificity directly from primary amino acid sequence information.
• Cofactor specificity prediction: Predicts specificity for flavin adenine dinucleotide (FAD(H2)), nicotinamide adenine dinucleotide (NAD(H)), and its phosphate form (NADP(H)).
• Rossmann fold identification: Identifies potential Rossmann fold cofactor-binding motifs within enzyme sequences.
• Algorithms: Employs Hidden Markov Models (HMMs) for sequence search and Artificial Neural Networks (ANNs) for cofactor specificity prediction.
• Training and benchmarking: Trained on experimental protein–cofactor structure complexes and benchmarked on an independent evaluation set with Matthews correlation coefficients of 0.94 for FAD(H2), 0.79 for NAD(H), and 0.65 for NADP(H).

Scientific Applications:

• Metabolic engineering: Enables optimization of cofactor balance in metabolically engineered microbial production strains to improve biosynthetic pathway performance.
• Heterologous enzyme selection: Facilitates identification of heterologous enzymes with altered cofactor requirements from native sequence content.
• Enzyme function studies: Supports biochemists and molecular biologists in predicting and manipulating enzyme cofactor usage from sequence data.

Methodology:

Hidden Markov Models (HMMs) are used for sequence searches to identify Rossmann folds; Artificial Neural Networks (ANNs) are used to predict cofactor specificity; training used experimental protein–cofactor structure complexes and benchmarking reported Matthews correlation coefficients of 0.94 (FAD(H2)), 0.79 (NAD(H)), and 0.65 (NADP(H)).