The identification of biomarker signatures in omics molecular profiling is usually performed to predict outcomes in a precision medicine context, such as patient disease susceptibility, diagnosis, prognosis, and treatment response. To identify these signatures, we have developed a biomarker discovery tool, called BioDiscML. From a collection of samples and their associated characteristics, i.e., the biomarkers (e.g., gene expression, protein levels, clinico-pathological data), BioDiscML exploits various feature selection procedures to produce signatures associated to machine learning models that will predict efficiently a specified outcome. To this purpose, BioDiscML uses a large variety of machine learning algorithms to select the best combination of biomarkers for predicting categorical or continuous outcomes from highly unbalanced datasets. The software has been implemented to automate all machine learning steps, including data pre-processing, feature selection, model selection, and performance evaluation. BioDiscML is delivered as a stand-alone program and is available for download at https://github.com/mickaelleclercq/BioDiscML.

Fast identification of microbial species in clinical samples is essential to provide an appropriate antibiotherapy to the patient and reduce the prescription of broad-spectrum antimicrobials leading to antibioresistances. MALDI-TOF-MS technology has become a tool of choice for microbial identification but has several drawbacks: it requires a long step of bacterial culture before analysis (≥24 h), has a low specificity and is not quantitative. We developed a new strategy for identifying bacterial species in urine using specific LC-MS/MS peptidic signatures. In the first training step, libraries of peptides are obtained on pure bacterial colonies in DDA mode, their detection in urine is then verified in DIA mode, followed by the use of machine learning classifiers (NaiveBayes, BayesNet and Hoeffding tree) to define a peptidic signature to distinguish each bacterial species from the others. Then, in the second step, this signature is monitored in unknown urine samples using targeted proteomics. This method, allowing bacterial identification in less than 4 h, has been applied to fifteen species representing 84% of all Urinary Tract Infections. More than 31,000 peptides in 190 samples were quantified by DIA and classified by machine learning to determine an 82 peptides signature and build a prediction model. This signature was validated for its use in routine using Parallel Reaction Monitoring on two different instruments. Linearity and reproducibility of the method were demonstrated as well as its accuracy on donor specimens. Within 4h and without bacterial culture, our method was able to predict the predominant bacteria infecting a sample in 97% of cases and 100% above the standard threshold. This work demonstrates the efficiency of our method for the rapid and specific identification of the bacterial species causing UTI and could be extended in the future to other biological specimens and to bacteria having specific virulence or resistance factors. Graphical Abstract Highlights Fast and culture-free method for the identification of the 15 bacterial species causing UTIs. Combination of DIA analysis and machine learning algorithms to define a peptide signature. High accuracy, good linearity and reproducibility, sensitivity below standard threshold. Transferability to other laboratories and other mass spectrometers.

Shotgun sequencing of environmental DNA (i.e., metagenomics) has revolutionized the field of environmental microbiology, allowing the characterization of all microorganisms in a sequencing experiment. To identify the microbes in terms of taxonomy and biological activity, the sequenced reads must necessarily be aligned on known microbial genomes/genes. However, current alignment methods are limited in terms of speed and can produce a significant number of false positives when detecting bacterial species or false negatives in specific cases (virus, plasmids, and gene detection). Moreover, recent advances in metagenomics have enabled the reconstruction of new genomes using de novo binning strategies, but these genomes, not yet fully characterized, are not used in classic approaches, whereas machine and deep learning methods can use them as models. In this article, we attempted to review the different methods and their efficiency to improve the annotation of metagenomic sequences. Deep learning models have reached the performance of the widely used k-mer alignment-based tools, with better accuracy in certain cases; however, they still must demonstrate their robustness across the variety of environmental samples and across the rapid expansion of accessible genomes in databases.

Secreted phospholipase A ₂ -IIA (sPLA ₂ -IIA) hydrolyzes phospholipids to liberate lysophospholipids and fatty acids. Given its poor activity toward eukaryotic cell membranes, its role in the generation of proinflammatory lipid mediators is unclear. Conversely, sPLA ₂ -IIA efficiently hydrolyzes bacterial membranes. Here, we show that sPLA ₂ -IIA affects the immune system by acting on the intestinal microbial flora. Using mice overexpressing transgene-driven human sPLA ₂ -IIA, we found that the intestinal microbiota was critical for both induction of an immune phenotype and promotion of inflammatory arthritis. The expression of sPLA ₂ -IIA led to alterations of the intestinal microbiota composition, but housing in a more stringent pathogen-free facility revealed that its expression could affect the immune system in the absence of changes to the composition of this flora. In contrast, untargeted lipidomic analysis focusing on bacteria-derived lipid mediators revealed that sPLA ₂ -IIA could profoundly alter the fecal lipidome. The data suggest that a singular protein, sPLA ₂ -IIA, produces systemic effects on the immune system through its activity on the microbiota and its lipidome.

Nutritional compounds may have an influence on different OMICs levels, including genomics, epigenomics, transcriptomics, proteomics, metabolomics, and metagenomics. The integration of OMICs data is challenging but may provide new knowledge to explain the mechanisms involved in the metabolism of nutrients and diseases. Traditional statistical analyses play an important role in description and data association; however, these statistical procedures are not sufficiently enough powered to interpret the large integrated multiple OMICs (multi-OMICS) datasets. Machine learning (ML) approaches can play a major role in the interpretation of multi-OMICS in nutrition research. Specifically, ML can be used for data mining, sample clustering, and classification to produce predictive models and algorithms for integration of multi-OMICs in response to dietary intake. The objective of this review was to investigate the strategies used for the analysis of multi-OMICs data in nutrition studies. Sixteen recent studies aimed to understand the association between dietary intake and multi-OMICs data are summarized. Multivariate analysis in multi-OMICs nutrition studies is used more commonly for analyses. Overall, as nutrition research incorporated multi-OMICs data, the use of novel approaches of analysis such as ML needs to complement the traditional statistical analyses to fully explain the impact of nutrition on health and disease.

Large language models (LLMs) originally developed for human text have been adapted to proteomics as protein language models (pLMs). These models treat amino acid sequences like sentences, and they learn patterns from millions of sequences. pLMs are used for several key tasks, including the prediction of protein structures, annotating protein functions, designing novel protein sequences with specific characteristics, and mapping the interactions between proteins and other molecules. Compared with traditional approaches, pLMs deliver insights more quickly but demand large computing resources and careful data management. Developers are focused on decreasing prediction inaccuracies and biases by exploring more efficient training techniques and smaller models to decrease the resources required. As sequence databases continue to grow, pLMs will improve to uncover links between proteins and disease pathways, speeding drug development and basic research while offering new proteome-scale insights that support experimental design and validation.