BioAutoMATED: An end-to-end automated machine learning tool for explanation and design of biological sequences

Physicians should build their own machine-learning models

“…Automated machine learning (AutoML) simplifies ML development and deployment by handling complex tasks like data processing and algorithm selection. 4 This approach aligns with my advocacy to make ML more accessible. Large technology companies now provide ML toolkits for users with limited technical experience in areas such as image analysis and language translation.…”

“…Moor et al.’s proposed Generalist Medical AI Framework 3 is an excellent demonstration. BioAutoMATED’s end-to-end automated ML tool for explanation and design of biological sequences 4 and OpenAI’s GPTs are both valid starting examples. Modular and accessible design: the final component enables healthcare professionals to assemble different parts of a ML model similar to using building blocks.…”

“…Moor et al.’s proposed Generalist Medical AI Framework 3 is an excellent demonstration. BioAutoMATED’s end-to-end automated ML tool for explanation and design of biological sequences 4 and OpenAI’s GPTs are both valid starting examples.…”

GPro: generative AI-empowered toolkit for promoter design

“…GPro provides implementations of most cutting-edge promoter design methods under the same framework, facilitating the analysis and comparison of diverse approaches and network architectures. Comparing with the general biological sequence designing tools like BioAutoMATED ( Valeri et al 2023 ), which utilize automated machine learning technology to construct models, GPro implements published state-of-the-art promoter design methods proposed by human expertise, aiming to enable users to gain a comprehensive understanding of the design process, and provide detailed guidance for selecting appropriate methods and creating their own pipeline. Our toolkit enables efficient in silico design of promoter sequences, and researchers without AI expertise can easily design promoters for their specific requirements.…”

Automated characterization and analysis of expression compatibility between regulatory sequences and metabolic genes in Escherichia coli

“…To investigate the influence of regulatory sequences and fusion protein coding sequences on fusion protein fluorescence values, we employed an end-to-end automated machine-learning approach using BioAutoMATED [ 30 ]. Based on the log 10 -transformated fluorescence values ( Supplementary Table S3 ), machine learning models were constructed for sequence data (comprising promoter, RBS, 180 bp of the gene, and (GGGGS) 3 linker alongside the GFP gene) and preprocessed fluorescence values using TOPT [ 31 ], DeepSwarm [ 32 ] and AutoKeras [ 33 ] architectures, with 85 % of the dataset allocated for training and 15 % for validation.…”