All articles tagged with #bioinformatics
AlphaFold’s database now includes 1.7 million predicted homodimer protein complexes, expanding its 200 million monomer structures to show how two copies of the same protein interact. The complex predictions were produced by a international consortium and are freely accessible via EMBL-EBI, broadening the resource’s biological and medical relevance across 20 studied species.
Evo 2 is a large, open biological foundation model trained on 9 trillion base pairs across all life, with a 1‑million‑token context, available in 7B and 40B sizes. It predicts mutational effects on DNA, RNA and proteins without task-specific fine-tuning, can generate genome-scale sequences and chromatin-accessibility designs, and supports exon–intron annotation and variant-effect prediction across diverse species. The model and data (OpenGenome2) are fully open-source, with training/inference code and tools (Evo Designer, Evo Mech Interp), plus safety safeguards such as virus-data exclusions to mitigate misuse.
Researchers at the University of Navarra have developed RNACOREX, an open-source software that maps gene regulation networks in cancer, helping to understand tumor behavior and predict patient survival with clear, interpretable results, advancing personalized cancer treatment.
The article introduces CellWhisperer, a multimodal AI framework that enables chat-based exploration and analysis of single-cell RNA-seq data using natural language, integrating transcriptome profiles with biological text understanding to facilitate intuitive data interrogation and hypothesis generation.
The article discusses MetaGraph, a scalable framework for indexing petabase-scale biological sequence repositories, enabling efficient and accurate full-text search across vast datasets like the SRA and ENA, with potential to revolutionize genomic research and data accessibility.
A new trimodal protein language model enhances the accuracy and efficiency of protein searches, leveraging advanced deep learning techniques to improve understanding of protein structures and functions.
LexicMap is a high-performance tool designed for efficient sequence alignment against millions of prokaryotic genomes, using a novel seeding approach with variable-length prefix and suffix matching, enabling rapid and scalable analysis suitable for large genomic databases.
Marine researchers discovered a new species of archaea in a mysterious black goo found inside a ship's rudder, revealing an unexpected habitat for extremophiles and potential applications in biofuel production, highlighting the importance of exploratory science.
Dimension, a venture capital firm, has successfully raised a $500 million second fund to invest in the intersection of biotechnology and machine learning, an area they describe as 'still contrary.' The firm, led by managing partners in New York and the Bay Area, aims to capitalize on the growing synergy between biological sciences and AI technologies.
Scientists have developed Evo, a machine learning model that predicts the effects of genetic mutations and generates new DNA sequences with high accuracy. Unlike traditional large language models, Evo is trained on microbial genomes, using base pairs as "words" to analyze genetic patterns. While Evo shows promise in understanding DNA and RNA functions, it currently lacks the ability to predict human genetic mutations. The researchers emphasize the importance of ethical guidelines to prevent misuse as the technology advances.
Researchers at UC San Diego have developed a "genetic atlas" using the model organism C. elegans to profile the function of nearly 500 genes during embryonic development. By blocking each gene one at a time and using time-lapse 4D imaging and computer vision, they tracked how these genes influence tissue formation and cell identity. This study, published in Cell, provides new insights into gene functions and their roles in development, with implications for understanding human developmental disorders. The data is now available through an online resource called PhenoBank.
The recent development of CRISPR-COPIES by the Center for Advanced Bioenergy and Bioproducts Innovation (CABBI) represents a significant advancement in genetic engineering, streamlining the identification of optimal genetic integration sites and accelerating the pace of scientific discovery and innovation. This high-tech upgrade to the CRISPR/Cas system offers a powerful and accessible tool for precision genome editing, with applications spanning agriculture, biofuel production, and gene therapy, promising to drive forward the boundaries of what’s possible in genetic engineering.
Researchers at KAUST have developed an innovative AI tool, DeepGO-SE, that excels in predicting the functions of unknown proteins, outperforming existing methods and ranking in the top 20 of over 1,600 algorithms in an international competition. Leveraging large language models and logical entailment, this tool can deduce molecular functions even for proteins without existing database matches, offering a groundbreaking approach to understanding cellular mechanisms. The team aims to apply this tool to explore proteins in extreme environments, opening new doors for biotechnological advancements.
A new AI tool called DeepGO-SE has been developed to predict the function of unknown proteins, outperforming existing methods and even analyzing proteins with no clear matches in existing datasets. Developed by KAUST bioinformatics researcher Maxat Kulmanov and colleagues, the tool employs logical entailment to draw meaningful conclusions about molecular functions based on general biological principles. It has been successful in predicting the functions of poorly understood proteins and is being used to investigate enigmatic proteins in plants thriving in extreme environments, with potential applications in drug discovery, disease associations, and more.
Scientists at Chapman University have developed a GenAI model called drugAI, which utilizes advanced AI techniques to design new drug compounds with the right properties and characteristics, promising to accelerate the process of identifying viable drug candidates for various diseases at a fraction of the cost. The model has been tested and validated, showing magnificent results and generating potential drugs that have never been conceived of before. It has also demonstrated a high validity rate, drug-likeness, and strong binding affinities to respective targets, making it a promising tool for future drug design and development.