Recep Adiyaman
Vision About Experience Education Skills Publications Tools Newsletter Updates Podcasts Blog Contact CV Subscribe
All Issues
Older Newer
Daily Signal February 16, 2026 · 8 min read

Issue #48: DeepFold-PLM: accelerating protein structure prediction via efficient homology search using protein language models.

Protein Design Digest - 2026-02-16 - DeepFold-PLM: accelerating protein structure prediction via efficient homology search using protein language models.

Share X LinkedIn
Protein Design Daily

Building something in Protein Design?

I love collaborating on new challenges. Let's build together.

LinkedIn Email

Subscribe to Protein Design Digest

Daily curated signals from arXiv, PubMed, and BioRxiv.

Signal of the Day

DeepFold-PLM: accelerating protein structure prediction via efficient homology search using protein language models.

Motivation Protein structure prediction has been revolutionized and generalized with the advent of cutting-edge AI methods such as AlphaFold, but reliance on computationally intensive multiple sequence alignments (MSA) remains a major limitation. Results We introduce DeepFold-PLM, a novel framework that integrates advanced protein language models with vector embedding databases to enhance ultra-fast MSA construction, remote homology detection, and protein structure prediction. DeepFold-PLM utilizes high-dimensional embeddings and contrastive learning, significantly accelerate MSA generation, achieving 47 times faster than standard methods, while maintaining prediction accuracy comparable to AlphaFold. In addition, it enhances structure prediction by extending modeling capabilities to multimeric protein complexes, provides a scalable PyTorch-based implementation for efficient large-scale prediction. Our method also effectively increases sequence diversity (Neff = 8.65 versus 4.83 with JackHMMER) enriching coevolutionary information critical for accurate structure prediction. DeepFold-PLM thus represents a versatile and practical resource that enables high-throughput applications in computational structural biology. Availability and implementation Source codes and user-friendly Python API of all modules of DeepFold-PLM publicly available at https://github.com/DeepFoldProtein/DeepFold-PLM.

Why this matters: Critical for improving fold accuracy and reducing structural uncertainty in de novo design.

Also Worth Reading

A New Insight into the Study of Neural Cell Adhesion Molecule (NCAM) Polysialylation Inhibition Incorporated the Molecular Docking Models into the NMR Spectroscopy of a Crucial Peptide-Ligand Interaction.

The expression of polysialic acid (polySia) on the neuronal cell adhesion molecule (NCAM) is called NCAM-polysialylation, which is strongly related to the migration and invasion of tumor cells and aggressive clinical status. During the NCAM polysialylation process, polysialyltransferases (polySTs), such as polysialyltransferase IV (ST8SIA4) or polysialyltransferase II (ST8SIA2), can catalyze the addition of CMP-sialic acid (CMP-Sia) to the NCAM to form polysialic acid (polySia). In this study, the docking models of polysialyltransferase IV (ST8Sia4) protein and different ligands were predicted using Alphafold 3 and DiffDock servers, and the prediction accuracy was further verified using the NMR experimental spectra of the interactions between polysialyltransferase domain (PSTD), a crucial peptide domain in ST8Sia4, and a different ligand. This combination strategy provides new insights into a quick and effective screening for inhibitors of tumor cell migration.

Benchmarking Generative AI Protein Models Reveals Differences Between Structural and Sequence-based Approaches.

Recent advances in artificial intelligence have led to the development of generative models for de novo protein design. We compared 13 state-of-the-art generative protein models, assessing their ability to produce feasible, diverse, and novel protein monomers. Structural diffusion models generally create designs with higher confidence in predicted structures and more biologically plausible energy distributions, but exhibit limited diversity and strong sequence biases. Conversely, protein language models generate more diverse and novel designs but with lower structural confidence. We also evaluated these models’ ability to generate unique proteins, conditionally based on the Tobacco Etch Virus (TEV) protease. Generative models were successful in producing functional enzymes, albeit with diminished activity compared to the wildtype TEV. Our systematic benchmarking provides a foundation for evaluating and selecting generative protein models, while highlighting the complementary strengths of different generative paradigms. This framework will facilitate an informed application of these tools for bio-medical engineering and design.

Investigation of the potential mechanism by which methylparaben induces psoriasis: an integrated study using network toxicology, molecular docking, molecular dynamics simulation, and eight machine learning algorithms.

Psoriasis is a chronic inflammatory skin disease with limited safe and effective treatments. Methylparaben, a widely used preservative in cosmetics, pharmaceuticals, and food, is an emerging environmental pollutant linked to immune-related skin disorders, but its role and mechanism in psoriasis remain unclear. This study explored its potential mechanism using network toxicology, molecular docking, molecular dynamics simulation, and eight machine learning algorithms. Methylparaben targets were retrieved from GeneCards and TCMSP, and psoriasis-related targets from CTD and GeneCards. Overlapping targets were screened with Venny 2.1.0. A PPI network was constructed via STRING, and core targets identified using Cytoscape 3.10.2. GO and KEGG enrichment analyses were performed on DAVID. Molecular docking evaluated the binding affinity of methylparaben with key targets. A total of 138 compound-related and 5,592 psoriasis-related targets were identified. Core targets such as INS, HIF1A, and PPARG are involved in regulating immune-inflammatory responses, keratinocyte proliferation and differentiation, and oxidative stress. GO analysis revealed enrichment in xenobiotic metabolism, lipopolysaccharide response, and metal ion binding. KEGG analysis highlighted pathways related to cancer, chemical carcinogenesis from reactive oxygen species, and drug metabolism via cytochrome P450 enzymes. Molecular docking showed stable binding of methylparaben to INS (-4.5 kcal/mol), HIF1A (-5.9 kcal/mol), and PPARG (-5.5 kcal/mol), primarily through hydrogen bonds and hydrophobic interactions. Methylparaben may exert its effects on psoriasis via multi-target and multi-pathway mechanisms, influencing inflammation, oxidative stress, and cellular regulation. These findings provide valuable insight into its toxicological mechanism and potential therapeutic application.

Research & AI Updates

From the Industry

Quick Reads

Demonstrating the Absence of Correlation Between Molecular Docking and in vitro Cytotoxicity in Anti-Breast Cancer Research: Root Causes and Practical Resolutions.

Introduction In silico methods have significantly transformed the landscape of drug discovery by enabling rapid and cost-effective screening of prospective therapeutic compounds. Read more →

From MM-PBSA to H-MMGB: Multiscale Modeling for Biomolecular Structure and Drug Discovery.

From early efforts to predict protein structure from simplified models, computational biophysics has progressed toward increasingly physics-based approaches for evaluating biomolecular structure, molecular interactions, and energetics. Read more →

Molecular docking approaches in mycetoma: Toward improved patient management.

Mycetoma is a neglected tropical disease characterised by chronic, granulomatous inflammation of the subcutaneous tissues, often leading to disfigurement, disability, and significant socioeconomic burdens. Read more →

Synthesis, biological evaluation, molecular docking, molecular dynamics simulation, and ADME studies of novel carbazole-aniline hybrids as cytotoxic agents.

A novel series of seven carbazole-aniline hybrids (5a-5 g) were designed, synthesized, and evaluated as potential anticancer agents. Read more →

Efficient global accuracy estimation for protein complex structural models using multi-view representation learning.

With the rapid advancement of protein structure prediction techniques and the explosive growth of predicted structural data, existing estimation of model accuracy (EMA) methods struggle to balance computational efficiency with estimation performance. Read more →

Detoxification mechanism of semi-coking wastewater by hydrogel-assisted SNAD process via iron‑mediated coordination adsorption: Performance, microbiota interaction, and molecular docking.

Semi-coking wastewater (SCWW) contains toxic compounds that threaten both human health and ecosystem integrity. Read more →

Linking spatial omics to patient phenotypes at the population scale by BSNMani: Bayesian scalar-on-network regression with manifold learning

Spatial omics enables the integration of high-dimensional molecular organization with clinical outcomes, yet incorporating spatial single-cell information into predictive models at the population scale remains challenging. Read more →

Kinic index: an artificial intelligence-driven predictive model and multitarget drug discovery framework for hepatocellular carcinoma patients.

Hepatocellular carcinoma (HCC) remains a major global health challenge due to its molecular heterogeneity, late diagnosis, and limited therapeutic options. Read more →

Pipeline Tip

Normalise thermal B-factors when comparing different crystal structures.

Resources & Tools

Deep learning is not a magic wand, but a powerful lens for structural biology. — Recep Adiyaman

BS HF DK