Exploration of bioinformatic domain based on data mining, reaction and enzyme promiscuity predictions

Alexei Lapkin,Chonghuan Zhang,Qianyue Zhang

DOI: https://doi.org/10.26434/chemrxiv-2023-913g7-v2

2024-10-03

Abstract:Biochemical transformations may allow significant improvements in synthetic efficiency of complex functional molecules through reduction in the number of synthetic steps or avoidance of harsh conditions and/or toxic solvents/reactants. Yet, there is a limited access to biochemical reaction data, which reduces the opportunities of finding alternatives and discovering synergies with organic synthesis. We propose a workflow to explore the sparse synthetic biological domain. Using a molecular graph method we predict feasible biosynthetic reactions. The products of biosyntheses are learned from the functional transformations of the literature-excerpted reactions recorded in KEGG database. Through this approach we expanded the KEGG reaction dataset of biochemical transformations by approximately four times. To catalyse the novel reactions, we proposed a transformer model that learns from reaction SMILES and amino acid sequences of native enzymes and predicts promiscuous enzymes for potential substrates. The proposed transformer model calibrates the feasibility of the predicted reactions and reduces the search scope for promiscuous enzymes in the pool. A populated biological reaction space is eventually visualised in a two-dimensional t-SNE diagram.

Chemistry

What problem does this paper attempt to address?

The paper attempts to address the issue of expanding biochemical reaction datasets to facilitate the application of synthetic biology in the synthesis of pharmaceuticals and industrial chemicals. Specifically: 1. **Data Scarcity Issue**: The existing biochemical reaction data is very limited, which restricts the opportunities to find alternative organic synthesis pathways and discover synergies with organic synthesis. 2. **Predicting New Reactions**: By extracting known biochemical reactions from the literature and using molecular graph methods to predict new biosynthetic reactions, the paper aims to expand the existing biochemical reaction database. 3. **Predicting Enzyme Multifunctionality**: To catalyze these newly predicted reactions, the paper proposes a transformer model based on natural language processing (NLP) that can learn from enzyme amino acid sequences and reaction SMILES, and predict multifunctional enzymes to catalyze potential substrate reactions. Through these methods, the researchers aim to enrich the biochemical reaction dataset and enhance the feasibility of synthetic biology in practical applications, particularly in reducing synthesis steps, avoiding harsh conditions, or toxic solvents/reactants when producing complex functional molecules.

Exploration of bioinformatic domain based on data mining, reaction and enzyme promiscuity predictions

Generation of connections between protein sequence space and chemical space to enable a predictive model for biocatalysis

Biocatalysed synthesis planning using data-driven learning

BCSExplorer: A Customized Biosynthetic Chemical Space Explorer with Multifunctional Objective Function Analysis.

Feasible-metabolic-pathway-exploration technique using chemical latent space

RxnBLAST: Molecular Scaffold and Reactive Chemical Environment Feature Extractor for Biochemical Reactions.

Metabolite and reaction inference based on enzyme specificities

Learning graph representations of biochemical networks and its application to enzymatic link prediction

ReactZyme: A Benchmark for Enzyme-Reaction Prediction

DORA-XGB: An improved enzymatic reaction feasibility classifier trained using a novel synthetic data approach

Predicting Metabolic Pathways of Small Molecules and Enzymes Based on Interaction Information of Chemicals and Proteins.

Predicting Enzymatic Reactions with a Molecular Transformer

Ideation and Evaluation of Novel Multicomponent Reactions via Mechanistic Network Analysis and Automation

Chemical reaction vector embeddings: towards predicting drug metabolism in the human gut microbiome

Machine-Learning-Guided Discovery of Electrochemical Reactions

Systematic Analysis of Enzyme-Catalyzed Reaction Patterns and Prediction of Microbial Biodegradation Pathways

Reaction performance prediction with an extrapolative and interpretable graph model based on chemical knowledge

An explainability framework for deep learning on chemical reactions exemplified by enzyme-catalysed reaction classification

Designing pathways for bioproducing complex chemicals by combining tools for pathway extraction and ranking