Biology subjects

Tagkopoulos, I.

Publications and source records attributed to Tagkopoulos, I..

4 recordsLinked to original sources

bioRxiv2024

FoodAtlas: Automated Knowledge Extraction of Food and Chemicals from Literature

Automated generation of knowledge graphs that accurately capture published information can help with knowledge organization and access, which have the potential to accelerate discovery and innovation. Here, we present an integrated pipeline to construct a large-scale knowledge graph using large language models in an active learning setting. We apply our pipeline to the association of raw food, ingredients, and chemicals, a domain that lacks such knowledge resources. By using an iterative active learning approach of 4,120 manually curated premise-hypothesis pairs as training data for ten consecutive cycles, the entailment model extracted 230,848 food-chemical composition relationships from 155,260 scientific papers, with 106,082 (46.0%) of them never been reported in any published database. To augment the knowledge incorporated in the knowledge graph, we further incorporated information from 5 external databases and ontology sources. We then applied a link prediction model to identify putative food-chemical relationships that were not part of the constructed knowledge graph. Validation of the 443 hypotheses generated by the link prediction model resulted in 355 new food-chemical relationships, while results show that the model score correlates well (R2 = 0.70) with the probability of a novel finding. This work demonstrates how automated learning from literature at scale can accelerate discovery and support practical applications through reproducible, evidence-based capture of latent interactions of diverse entities, such as food and chemicals.

bioinformatics↗

bioRxiv2022

Machine learning prediction models for nutrient profiles in food after processing

The future of personalized health relies on knowledge of dietary composition. The current analytical methods are impractical to scale up, and the computational methods are inadequate. We propose machine learning models to predict the nutritional profiles of cooked foods given the raw food composition and cooking method, for a variety of plant and animal-based foods. Our models (trained on USDAs SR dataset) were on average 31% better than baselines, based on RMSE metric, and particularly good for leafy green vegetables and various cuts of beef. We also identified and remedied a bias in the data caused by representation of composition per 100grams. The scaling methods are based on a process-invariant nutrient, and the scaled data improves prediction performance. Finally, we advocate for an integrated approach of data analysis and modeling when generating future composition data to make the task more efficient, less costly and apply for development of reliable models.

bioinformatics↗

bioRxiv2021

The Atomic-Level Physiochemical Determinants of T Cell Receptor Dissociation Kinetics

The rational design of T Cell Receptors (TCRs) for immunotherapy has stagnated due to a limited understanding of the dynamic physiochemical features of the TCR that elicit an immunogenic response. The physiochemical features of the TCR-peptide major histocompatibility complex (pMHC) bond dictate bond lifetime which, in turn, correlates with immunogenicity. Here, we: i) characterize the force-dependent dissociation kinetics of the bond between a TCR and a set of pMHC ligands using Steered Molecular Dynamics (SMD); and ii) implement a machine learning algorithm to identify which physiochemical features of the TCR govern dissociation kinetics. Our results demonstrate that the total number of hydrogen bonds between the CDR2{beta}-MHC({beta}), CDR1-Peptide, and CDR3{beta}-Peptide are critical features that determine bond lifetime. We propose that amino acid substitutions to these hypervariable regions of the TCR can efficiently manipulate immunogenicity and thus be used in the rational design of TCRs for immunotherapy.

immunology↗

bioRxiv2018

A network-based model for drug repurposing in Rheumatoid Arthritis

BackgroundThe identification of drug repositioning targets through computational methods has the potential to provide a fast, inexpensive alternative to traditional drug discovery process. Diseases where complicated pathophysiology evolves over time are excellent targets for such methods. One such disease is Rheumatoid Arthritis (RA), a chronic inflammatory autoimmune disease, where the drug survival at 5-years is less than 50% for patients treated with disease-modifying anti-rheumatic drugs (DMARDs).\n\nMethods and FindingsWe have developed a network-based approach for drug repositioning that takes into account the human interactome network, proximity measures between drug targets and disease-associated genes, potential side-effects, genome-wide gene expression and disease modules that emerge through pertinent analysis. We found that all DMARDs, except for hydroxychloroquine (HCQ), were found to be significantly proximal to RA-related genes. Application of the method on anti-diabetic agents, statins and H2 receptor blockers identified anti-diabetic agents - gliclazide, sitagliptin and metformin - that have similar network signatures with the DMARDs. Subsequent in-vitro experiments on mouse fibroblast NIH-3T3 cells validated the findings and the down regulation of six key RA-related inflammatory genes. Our analysis further argues that the combination of HCQ and/or sulfasalazine with methotrexate (MTX) is predicted to have an additive synergistic effect in treatment based on network complementarity. Similarly, leflunomide and tofacitinib were found to be suitable alternatives upon chemoresistance to MTX-based double/triple therapy, given the complementary network signatures and overlapping critical target hubs.\n\nConclusionsOur results corroborate that computational methods that are based on network proximity, among other contextual information can help narrow down the drug candidates for drug repositioning, as well as support decisions for combinatorial drug treatment that is tailored to patients needs.\n\nAuthor summaryThe network-based proximity between drug targets and disease genes can provide novel insights on the repercussion, interplay, and reposition of drugs in the context of disease. Disease-modifying anti-rheumatic drugs (DMARDs) located significantly close to rheumatoid arthritis (RA)-associated genes and RA-relevant pathways. Three anti-diabetic agents were identified to have an anti-inflammatory effect like DMARDs. We built RA disease module encompassing the emerging small-molecule targets and functional neighbors, which better explained the RA pathophysiology. By proximity and network robustness, tofacitinib and tocilizumab were the most potent for RA disease module as a single agent, and this is consistent with clinical observation. Side effects of clinical importance are predictable by measuring network-based proximity between drug targets and side effect protein. Network-based drug-disease proximity offers a novel and clinically actionable information about drugs and opens the new possibility to drug combination and reposition.

systems biology↗