PRONTO-TK: a user-friendly PROtein Neural neTwOrk tool-kit for accessible protein function prediction

NAR Genomics and Bioinformatics, Journal Year: 2024, Volume and Issue: 6(3)

Published: July 2, 2024

Abstract Associating one or more Gene Ontology (GO) terms to a protein means making statement about particular functional characteristic of the protein. This association provides scientists with snapshot biological context activity. paper introduces PRONTO-TK, Python-based software toolkit designed democratize access Neural-Network based complex function prediction workflows. PRONTO-TK is user-friendly graphical interface (GUI) for empowering researchers, even those minimal programming experience, leverage state-of-the-art Deep Learning architectures annotation using GO terms. We demonstrate PRONTO-TK’s effectiveness on running example, by showing how its intuitive configuration allows it easily generate analyses while avoiding complexities building such pipeline from scratch.

Language: Английский

---

Improving Prediction of Complications Post-Proton Therapy in Lung Cancer Using Large Language Models and Meta-Analysis

Cancer Control, Journal Year: 2024, Volume and Issue: 31

Published: Jan. 1, 2024

This study enhances the efficiency of predicting complications in lung cancer patients receiving proton therapy by utilizing large language models (LLMs) and meta-analytical techniques for literature quality assessment.

Language: Английский

---

Integrating genetic algorithms and language models for enhanced enzyme design

Briefings in Bioinformatics, Journal Year: 2024, Volume and Issue: 26(1)

Published: Nov. 22, 2024

Abstract Enzymes are molecular machines optimized by nature to allow otherwise impossible chemical processes occur. Their design is a challenging task due the complexity of protein space and intricate relationships between sequence, structure, function. Recently, large language models (LLMs) have emerged as powerful tools for modeling analyzing biological sequences, but their application limited high cardinality space. This study introduces framework that combines LLMs with genetic algorithms (GAs) optimize enzymes. trained on dataset sequences learn amino acid residues linked structure knowledge then leveraged GAs efficiently search improved catalytic performance. We focused two optimization tasks: improving feasibility biochemical reactions increasing turnover rate. Systematic evaluations 105 biocatalytic demonstrated LLM–GA generated mutants outperforming wild-type enzymes in terms 90% instances. Further in-depth evaluation seven reveals power this methodology make “the best both worlds” create structural features flexibility comparable wild types. Our approach advances state-of-the-art computational biocatalysts, ultimately opening opportunities more sustainable processes.

Language: Английский

---

From Pre-Training to Meta-Learning: A Journey in Low-Resource-Language Representation Learning

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 115951 - 115967

Published: Jan. 1, 2023

Language representation learning is a vital field in Natural Processing (NLP) that aims to capture the intricate semantics and contextual information of text. With advent deep neural network architectures, has revolutionized NLP landscape. However, majority research this concentrated on resource-rich languages, putting Low-Resource Languages (LRL) at disadvantage due their limited linguistic resources absence pre-trained models. This paper addresses significance language low-resource language, Greek, its impact various downstream tasks heavily rely semantically contextually enriched representations. Accurate classification requires an understanding nuanced cues dependencies. Effective representations bridge gap between raw text data models, encoding semantic meaning, syntactic structures, information. By leveraging techniques using Transformer-based Models (LM), such as domain-adaption contrastive learning, we aim enhance performance LRL setting. We explore challenges opportunities developing effective propose multi-stage LM pre-training meta-learning approach improve tasks. The proposed was evaluated Greek expert-annotated texts from social media posts, news articles, press clippings internet articles blog posts opinion pieces. results show significant improvements effectiveness each task terms micro-averaged F1-score sentiment, irony, hate speech, emotion three custom

Language: Английский

---

Development of method using language processing techniques for extracting information on drug–health food product interactions

British Journal of Clinical Pharmacology, Journal Year: 2024, Volume and Issue: 90(6), P. 1514 - 1524

Published: March 20, 2024

Health food products (HFPs) are foods and related to maintaining promoting health. HFPs may sometimes cause unforeseen adverse health effects by interacting with drugs. Considering the importance of information on interactions between drugs, this study aimed establish a workflow extract Drug-HFP Interactions (DHIs) from open resources.

Language: Английский

---