Graph neural networks for conditional de novo drug design

🔬 What are GNNs?
GNNs are a type of deep learning model that work with data structured as graphs. Unlike traditional neural networks, GNNs can handle complex data, like molecular structures, social networks, or maps.

🧬When combined with conditioning techniques, GNNs can generate molecules with desired chemical and biological properties, opening new possibilities for drug design.

The paper explores how graph-based molecular representation and various GNN architectures are used in drug design. It also discusses methods for conditional generation to optimize molecular properties, while highlighting key challenges and future directions for improving the application of GNNs in drug discovery.

GNN-powered tools holds immense promise for making drug discovery more efficient and innovative.

Read the full length paper

#ResourcePost #GNN #DrugDiscovery

Stay Connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

📢 Using Residue Interaction Networks to understand protein function and evolution and to engineer new proteins

Residue Interaction Networks (RINs) are a powerful tool for protein analysis

- RINs offer a graph-based representation of protein structures.
- RINs are computationally efficient and have been widely used to explore protein stability, function, and allostery.
- RINs have been successfully applied in protein folding, stability, and interaction studies, offering a framework for creating designer proteins.

The paper highlights various tools available for RIN analysis, which handle different types of interactions and inputs, such as crystal structures, simulation data, and protein comparisons.

Learn More

#ResourcePost #RIN #ProteinEngineering


Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

📢 Artificial intelligence tools for the identification of antibiotic resistance genes

- AI tools are emerging as a solution to overcome limitations in traditional antibiotic resistance gene (ARG) detection methods.
- Techniques like machine learning and deep learning can directly classify ARGs from genome and plasmid sequences.
- AI can help identify novel ARGs and predict potential antibiotic targets, aiding in the fight against antibiotic resistance.

The review explains how AI-based methods can enhance the detection and annotation of ARGs. It explores the current AI techniques used in bioinformatics to address the growing threat of antibiotic resistance by identifying new ARGs and improving our understanding of bacterial defense mechanisms.

Learn more

#ResourcePost #AMR #ARG

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

📢 CellChat for systematic analysis of cell–cell communication from single-cell transcriptomics

Recent advances in single-cell sequencing technologies have revolutionized the way we study cell-cell communication.
CellChat is a new computational tool that offers an intuitive and systematic framework for analyzing communication networks between cells using single-cell transcriptomic data.

CellChat uses a simplified mass-action-based model to quantify signaling communication probabilities between cell groups, incorporating interactions between ligands, receptors, and cofactors. The tool is available in R, requires basic knowledge of single-cell data analysis, and offers a user-friendly interface for biologists without specialized bioinformatics training.

Learn more

#ResourcePost #CellChat

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

The World Health Organization ranks antibiotic resistance as one of the top 3 threats to global health. To combat this, researchers have developed a new family of antibiotics targeting the enzyme AcpS, crucial for bacterial survival

🧬 What makes this exciting os that no current antibiotics target AcpS! 🚀

🔬 Key Discoveries:
- A library of 700+ novel compounds, with 33 showing strong bacterial inhibition at low doses (≤ 2 μg/mL).
- Effective against Gram-positive bacteria and combined with colistin to tackle Gram-negative species.
- Success against multi-drug resistant strains in lab tests and animal models, including tough-to-treat infections like diabetic foot ulcers.

This innovative antibiotic family could be the foundation for new treatments against superbugs.

Learn more

#ResourcePost #AMR

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

✨️Evaluating large language model workflows in clinical decision support: referral, triage, and diagnosis


Accurate medical decisions are crucial for both patients and clinicians. Patients often struggle to interpret symptoms and choose the right specialist, while clinicians face challenges in processing complex patient data .

Recent advancements in large language models (LLMs) are transforming how we support decision-making in healthcare 🤖. In this study, the researchers benchmarked several LLM versions and an LLM-based workflow using retrieval-augmented generation (RAG) on 2,000 medical cases from the Medical Information Mart for Intensive Care database.

These LLMs hold great promise for improving patient outcomes and streamlining healthcare processes! 🌟

Learn more

#ResourcePost #LLM

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

🧬 BindCraft: one-shot design of functional protein binders

Protein-protein interactions (PPIs) are at the heart of all biological processes, but designing them has always been a complex challenge... until now! 🎉

Meet BindCraft – an open-source, automated pipeline for de novo protein binder design with experimental success rates between 10-100%. Powered by AlphaFold2 deep learning models, BindCraft generates nanomolar binders without the need for high-throughput screening or experimental optimization, even when binding sites are unknown.

This advancement brings us closer to a "one design-one binder" approach, opening doors to new possibilities in therapeutics, diagnostics, and biotechnology 🧪.

Learn more

#ResourcePost #PPIs

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

📢 Preliminary study on cytotoxicity of selegiline on different cancer cell lines: exploration of the induction of ROS-independent apoptosis in breast cancer cells

This study explores the cytotoxic effects of Selegiline on various cancer cell lines, with a focus on breast cancer cells. Interestingly, the research highlights ROS-independent apoptosis indicating that Selegiline may induce cell death in breast cancer cells without relying on reactive oxygen species (ROS). This could open new avenues for targeted cancer therapies.

Learn more

#ResourcePost #CancerResearch

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

📢 Modeling Protein–Glycan Interactions with HADDOCK

The study explores the use of High Ambiguity-Driven Docking (HADDOCK) to predict protein–glycan complexes, which are crucial for understanding biological signaling and functions related to glycans.

Glycans, molecules made of monosaccharides, interact with proteins for information transfer and signaling, and their structural diversity is key to these processes. The study highlights the importance of determining the 3D structures of protein–glycan complexes for both biological understanding and drug design.

HADDOCK is a molecular docking tool that was tested on 89 protein–glycan complexes, yielding a 70% success rate for bound complexes and a 40% success rate for unbound forms when some knowledge of the binding site was assumed.

#ResourcePost #Haddock

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

🌐 Vitessce: integrative visualization of multimodal and spatially resolved single-cell data

Vitessce is an interactive, web-based framework designed to tackle the challenge of visualizing high-dimensional single-cell and spatially resolved multiomics data. It allows researchers to explore a vast array of data types, including:

- Transcriptomic, proteomic, and genome-mapped data
- Cell-type annotations and gene expression profiles
- Spatially resolved transcripts and cell segmentations

Vitessce integrates millions of data points into coordinated views, providing a comprehensive solution for multiomics analysis. Best of all, it's open-source and ready to enhance your research today!

Learn more

#ResourcePost #Vitessce

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

🚨Predicting the Dynamic Interaction of Intrinsically Disordered Proteins

📊 GSALIDP is a GraphSAGE-embedded LSTM network designed to predict dynamic interactions of Intrinsically Disordered Proteins (IDPs).

📈 It models IDP conformations as dynamic graphs, using data from atomistic molecular dynamics (MD) simulations.

🔍 The model effectively predicts interaction sites and contact residue pairs, outperforming traditional models used for structural proteins.

🚀 This is a first of its kind method for predicting IDP-involved interactions!

Learn more

#ResourcePost #IDP

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

💊Enhancing Biomarker-Based Oncology Trial Matching Using Large Language Models

Clinical trials are crucial for developing new cancer treatments, yet determining patient eligibility is often hindered by the vast amount of unstructured data.

🧬 Genomic biomarkers play a critical role in precision medicine, making them essential for matching patients to appropriate trials.

📊 This study examines the use of Large Language Models (LLMs) to extract genomic biomarkers from oncology trial data, aiming to improve patient enrollment. Results indicate that open-source models outperform closed-source models, such as GPT-4, and demonstrate even greater accuracy when fine-tuned with additional data.

Learn more

#ResourcePost #LLM #CancerResearch

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

📢 Design of linear and cyclic peptide binders of different lengths from protein sequence information

A new method, EvoBind2, offers a groundbreaking approach to designing linear and cyclic peptide binders based on a given protein target sequence. Unlike traditional methods, EvoBind2 does not require predefined binding sites or binder lengths, making it a completely blind design process.

🔬 Key Highlights:
- Supports design of both linear and cyclic peptides of varying lengths
- Achieves nM affinity in a single shot
- Avoids adversarial designs through orthogonal in silico evaluation, tripling the success rate
- Addresses the challenge of peptide flexibility with an innovative strategy

Learn more

#ResourcePost #EvoBind2 #Protein


Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

📢 Unmasking AlphaFold to integrate experiments and predictions in multimeric complexes

AF_unmasked tackles the limitations of AlphaFold by integrating experimental data to predict large or hard-to-determine protein complexes with high confidence.

🔬 Key Highlights
- Builds larger assemblies with DockQ score > 0.8
- Fills incomplete experimental structures (structural inpainting), offering insights into protein dynamics
- Works even with limited evolutionary data or imperfect experimental structures
- Bridges experimental data and prediction pipelines for multimeric protein structures

AF_unmasked offers an easy-to-use approach to enhance structural predictions and support experimental research.

Learn more

#ResourcePost #AlphaFold

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

🚨 Enhancing heart disease classification with M2MASC and CNN-BiLSTM integration for improved accuracy🫀

Introducing the M2MASC-enabled CNN-BiLSTM model, integrating blockchain and IoT for real-time, secure health data analysis 🌐🔒.

Key Features:
- VGG16 pre-trained model for enhanced feature extraction 🎯
- Blockchain ensures data security, transparency, and immutability 🔑
- Real-time data from IoT devices for continuous patient monitoring 📡
- Optimized using mixed attention and search algorithms for top-tier performance 🛠️

Performance Metrics:
✅ Accuracy: 98.25%
✅ Precision: 99.57%
✅ Recall: 97.53%

Tested on the MIT-BIH dataset, this model promises to outperform traditional methods in predictive accuracy and data integrity 📊.

Learn more

#ResourcePost #Bioinformatics

📌 Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

🌍 Using artificial intelligence to document the hidden RNA virosphere

Scientists have developed LucaProt, a powerful deep learning tool that dives into 10,487 metatranscriptomes from ecosystems worldwide to uncover highly divergent RNA viruses. 🚀

✨ Key Discoveries:
- 🦠 161,979 potential RNA virus species & 180 new supergroups identified!
- 📏 Unveiled giant genomes up to 47,250 nucleotides long
- 🔬 RNA viruses confirmed through RT-PCR & sequencing
- 🌋 Found in diverse environments like air, hot springs, and hydrothermal vents

This groundbreaking tool is opening new doors to understanding the global RNA virome and its hidden diversity!

Learn more

#ResourcePost #RNA #Virus

📌 Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

📢 MetaLab Platform Enables Comprehensive DDA and DIA Metaproteomics Analysis


Metaproteomics delves into the protein composition of microbial communities, offering critical insights into their roles across ecosystems. However, the sheer size and diversity of metaproteomic data pose significant challenges for analysis.

This paper presents an updated MetaLab software, revolutionizing metaproteomic analysis with key innovations:

- Supports both DDA & DIA modes across MS platforms (Orbitrap, timsTOF)
- DIA analysis without DDA results, providing greater experimental flexibility
- Deep learning integration for improved peptide identification accuracy

MetaLab’s versatility makes it an essential tool, streamlining metaproteomic analysis and empowering researchers to better understand microbial communities' functions and dynamics.

Learn more

#ResourcePost #Metaproteomics

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

💥 PROPERMAB: an integrative framework for in silico prediction of antibody developability using machine learning

The new computational framework PROPERMAB is revolutionizing antibody development!
By leveraging machine learning and custom molecular features, it predicts key developability properties such as Hydrophobic interaction chromatography retention time and High-concentration viscosity

With PROPERMAB, researchers can now assess antibody developability early in the process, reducing timelines and saving resources ⚡. This framework scales effortlessly to large sequence datasets, enabling rapid, high-throughput predictions for therapeutic candidates.

Learn more

#ResourcePost #AntibodyDevelopment

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum

Automated machine learning with interpretation: A systematic review of methodologies and applications in healthcare

A recent systematic review covers the evolving role of Automated Machine Learning (AutoML) in healthcare, especially with interpretation systems that make ML models more accessible and understandable for clinicians and researchers. Examining 118 studies, this resource offers insights into key AutoML techniques, including automated data preparation, feature engineering, and model development, along with interpretation methods that enhance trust and usability in clinical settings.

This review serves as a valuable resource for professionals aiming to leverage AutoML’s potential in healthcare, promoting innovation in data-driven decision-making

Learn more

#ResourcePost #MachineLearning

Stay connected with us:
📢 Channel
🔵 LinkedIn
💬 WhatsApp
👍 Facebook
📸 Instagram
📣 Discussion Forum