Omic data analysis is steadily growing as a driver of basic and applied molecular biology research. Core to the interpretation of complex and heterogeneous biological phenotypes are computational approaches in the fields of statistics and machine learning. In parallel, constraint-based metabolic modeling has established itself as the main tool to investigate large-scale relationships between genotype, phenotype, and environment. The development and application of these methodological frameworks have occurred independently for the most part, whereas the potential of their integration for biological, biomedical, and biotechnological research is less known. Here, we describe how machine learning and constraint-based modeling can be combined, reviewing recent works at the intersection of both domains and discussing the mathematical and practical aspects involved. We overlap systematic classifications from both frameworks, making them accessible to nonexperts. Finally, we delineate potential future scenarios, propose new joint theoretical frameworks, and suggest concrete points of investigation for this joint subfield. A multiview approach merging experimental and knowledge-driven omic data through machine learning methods can incorporate key mechanistic information in an otherwise biologically-agnostic learning process.

Significance: Linking genotype and phenotype is a fundamental problem in biology, key to several biomedical and biotechnological applications. Cell growth is a central phenotypic trait, resulting from interactions between environment, gene regulation, and metabolism, yet its functional bases are still not completely understood. We propose and test a machine-learning approach that integrates large-scale gene expression profiles and mechanistic metabolic models, for characterizing cell growth and understanding its driving mechanisms in Saccharomyces cerevisiae . At its core, a custom-built multimodal learning method merges experimentally generated and model-generated data. We show that our approach can leverage the advantages of both machine learning and metabolic modeling, revealing unknown interactions between biological domains, incorporating mechanistic knowledge, and therefore overcoming black-box limitations of conventional data-driven approaches.

Gene regulation is responsible for controlling numerous physiological functions and dynamically responding to environmental fluctuations. Reconstructing the human network of gene regulatory interactions is thus paramount to understanding the cell functional organization across cell types, as well as to elucidating pathogenic processes and identifying molecular drug targets. Although significant effort has been devoted towards this direction, existing computational methods mainly rely on gene expression levels, possibly ignoring the information conveyed by mechanistic biochemical knowledge. Moreover, except for a few recent attempts, most of the existing approaches only consider the information of the organism under analysis, without exploiting the information of related model organisms. We propose a novel method for the reconstruction of the human gene regulatory network, based on a transfer learning strategy that synergically exploits information from human and mouse, conveyed by gene-related metabolic features generated in silico from gene expression data. Specifically, we learn a predictive model from metabolic activity inferred via tissue-specific metabolic modelling of artificial gene knockouts. Our experiments show that the combination of our transfer learning approach with the constructed metabolic features provides a significant advantage in terms of reconstruction accuracy, as well as additional clues on the contribution of each constructed metabolic feature. The method, the datasets and all the results obtained in this study are available at: https://doi.org/10.6084/m9.figshare.c.5237687. Supplementary data are available at Bioinformatics online.

Background: The rapid accumulation of sequencing data from metagenomic studies is enabling the generation of huge collections of microbial genomes, with new challenges for mapping their functional potential. In particular, metagenome-assembled genomes are typically incomplete and harbor partial gene sequences that can limit their annotation from traditional tools. New scalable solutions are thus needed to facilitate the evaluation of functional potential in microbial genomes. Methods: To resolve annotation gaps in microbial genomes, we developed KEMET, an open-source Python library devised for the analysis of Kyoto Encyclopedia of Genes and Genomes (KEGG) functional units. KEMET focuses on the in-depth analysis of metabolic reaction networks to identify missing orthologs through hidden Markov model profiles. Results: We evaluate the potential of KEMET for expanding functional annotations by simulating the effect of assembly issues on real gene sequences and showing that our approach can identify missing KEGG orthologs. Additionally, we show that recovered gene annotations can sensibly increase the quality of draft genome-scale metabolic models obtained from metagenome-assembled genomes, in some cases reaching the accuracy of models generated from complete genomes. Conclusions: KEMET therefore allows expanding genome annotations by targeted searches for orthologous sequences, enabling a better qualitative and quantitative assessment of metabolic capabilities in novel microbial organisms.

Background: The anaerobic digestion process degrades organic matter into simpler compounds and occurs in strictly anaerobic and microaerophilic environments. The process is carried out by a diverse community of microorganisms where each species has a unique role and it has relevant biotechnological applications since it is used for biogas production. Some aspects of the microbiome, including its interaction with phages, remains still unclear: a better comprehension of the community composition and role of each species is crucial for a cured understanding of the carbon cycle in anaerobic systems and improving biogas production. Results: The primary objective of this study was to expand our understanding on the anaerobic digestion microbiome by jointly analyzing its prokaryotic and viral components. By integrating 192 additional datasets into a previous metagenomic database, the binning process generated 11,831 metagenome-assembled genomes from 314 metagenome samples published between 2014 and 2022, belonging to 4,568 non-redundant species based on ANI calculation and quality verification. CRISPR analysis on these genomes identified 76 archaeal genomes with active phage interactions. Moreover, single-nucleotide variants further pointed to archaea as the most critical members of the community. Among the MAGs, two methanogenic archaea, Methanothrix sp. 43zhSC_152 and Methanoculleus sp. 52maCN_3230, had the highest number of SNVs, with the latter having almost double the density of most other MAGs. Conclusions: This study offers a more comprehensive understanding of microbial community structures that thrive at different temperatures. The findings revealed that the fraction of archaeal species characterized at the genome level and reported in public databases is higher than that of bacteria, although still quite limited. The identification of shared spacers between phages and microbes implies a history of phage-bacterial interactions, and specifically lysogenic infections. A significant number of SNVs were identified, primarily comprising synonymous and nonsynonymous variants. Together, the findings indicate that methanogenic archaea are subject to intense selective pressure and suggest that genomic variants play a critical role in the anaerobic digestion process. Overall, this study provides a more balanced and diverse representation of the anaerobic digestion microbiota in terms of geographic location, temperature range and feedstock utilization.