Biology subjects

Mirdita, M.

Publications and source records attributed to Mirdita, M..

At least 19 recordsLinked to original sources

bioRxiv2026

Systematic exploration of predicted quaternary structures within pandemic-relevant viral proteomes

Mechanistic understanding of viral protein-protein interactions enables global health security and pandemic preparedness, yet experimental characterisation remains difficult, costly, and often restricted to specialised laboratories. Here, we describe an in silico campaign using AlphaFold2 and AlphaFold-Multimer to predict monomers and dimer structures within 2,812 viral proteomes from 23 viral families relevant to human health. We report high-confidence predicted structures of 5,279 hetero-, and 2,749 homo-dimers. Structural clustering of interfaces reduces the set fivefold to 1,598 consolidated by form and function, of which 471 (29.5%) lacked detectable similarity to experimentally determined interfaces in the PDB. These clusters yielded two notable findings: new data on vaccinology-relevant glycoproteins, and accurate prediction of the specificity and cleavage-site recognition of viral proteases, which are targets for anti-viral development. Viral monomers and high-confidence dimers are available for interactive browsing in the Pandemic Preparedness Portal of the AlphaFold Database (https://alphafold.ebi.ac.uk/), and the full dataset is available for bulk download.

bioinformatics↗

bioRxiv2026

BFVD v3-UniProt-complete, improved viral protein structure predictions

The Big Fantastic Virus Database (BFVD) v3 is the most comprehensive resource for ColabFold-AlphaFold2-predicted viral protein structures. It holds 5,776,417 structures of nearly all viral sequences in UniProt 2025_03, a 16.4-fold increase compared to the representative-only catalogs of BFVD v1 and v2. Structure prediction quality has also improved, with high-confidence predictions accounting for 75.3% of BFVD v3 entries. This is due to Logan's enormous sequence assembly, now mined for the entire BFVD v3 instead of only for shallow alignments, alongside continued prediction quality improvements. BFVD v3 covers 72.7% of ICTV's virus species, 1.54-fold more than BFVD v2. With this version, the fraction of fully covered viral reference proteomes increases from 1.5% to 72.6%. BFVD v3 thus brings us closer to a structural catalog of the known virosphere and is freely available at bfvd.steineggerlab.workers.dev and bfvd.foldseek.com.

bioinformatics↗

bioRxiv2026

Fast remote nucleotide sequence alignment with Riboseek

Structure prediction has reached RNA, where generating deep alignments is now the bottleneck. We developed Riboseek, a search and alignment tool for RNA and DNA that represents sequences as overlapping di-mers. Riboseek is more sensitive for homology detection than BLASTN and nhmmer, and generates alignments 250- and 376-fold faster than nhmmer and rMSA. With structure-aware realignment, it approaches rMSAs base-pair recovery, while remaining 29-fold faster. We provide a web server and 1.7 million precomputed alignments for putatively structured RNAs. Riboseek is freely available at https://github.com/steineggerlab/riboseek.

bioinformatics↗

bioRxiv2026

AlphaFold Database expands to proteome-scale quaternary structures

Protein function is governed by molecular interactions, yet structural coverage of these interactions remains sparse. The AlphaFold Protein Structure Database (AFDB) transformed access to accurate monomeric protein structures at scale. Here, we expand the AFDB to quaternary structures by predicting 31M candidate homo- and heterodimeric protein complexes, compiled from 4,777 proteomes, including model- and global health organisms. We established confidence criteria via analysis of experimentally determined structures, resulting in 1.81M high-confidence predictions. These models enabled the discovery of emergent structures and topologies not present in monomeric predictions. Additionally, the top 1% of structural clusters accounted for [~]44% of all complexes, and [~]8.3% of clusters were conserved across multiple domains of life, pointing to a substantial fraction of ancient, universally retained assemblies. Structural search highlighted that high-confidence predictions are anchored in experimental multimer space, yet 31.3% extend beyond detectable PDB coverage. These freely accessible proteome-scale predictions facilitate functional and mechanistic hypothesis generation across biology.

synthetic biology↗

bioRxiv2026

evedesign: accessible biosequence design with a unified framework

Machine learning methods for protein engineering are rarely interoperable, require bespoke workflows, and remain inaccessible to non-experts. Yet the design problems that matter most - conditional design subject to real-world constraints, multi-objective optimization, and iterative lab-in-the-loop workflows where experimental data continuously refines successive design rounds - demand exactly the kind of flexible, composable infrastructure that no single tool provides. We present evedesign, a unified open-source framework that formalizes conditional biosequence design in a method-agnostic way, enabling complex multiobjective workflows combining supervised and unsupervised models from standardized specifications, and built from the outset to support iterative experimental integration. An interactive web interface facilitates end-to-end design for a broad scientific audience at https://evedesign.bio. We demonstrate evedesigns utility in antibody engineering, enzyme design, and natural enzyme discovery, and invite open-source community contributions.

bioinformatics↗

bioRxiv2026

Efficient protein structure prediction fromcompact computers to datacenters withOpenFold-TRT

We introduce accelerations for deep learning inference with OpenFold and TensorRT that, combined with MMseqs2-GPU on an x86 system with one NVIDIA RTX PRO 6000 Blackwell Server Edition GPU, reach up to 131x faster inference compared to AlphaFold2. ARM-optimizations enable homology search on power efficient DGX Spark, and allow search beyond available GPU RAM on NVIDIA Grace Hopper Superchip at comparable overall folding speed. These accelerations enable high-throughput protein structure inference at no accuracy cost using familiar tools.

bioinformatics↗

bioRxiv2026

Protein Language Modeling beyond static folds reveals sequence-encoded flexibility

MotivationProteins function through motion. Yet, most discoveries still commence with static representations of protein structures. Here, we investigated the feasibility of leveraging protein dynamics to improve homology detection. ResultsWe introduce ProtProfileMD, a sequence-to-3D-probability model that predicts, from an amino acid sequence, a profile of discrete structural representations capturing protein dynamics. We applied supervised parameter-efficient finetuning of the ProstT5 protein Language Model (pLM) to predict per-residue distributions over Foldseeks 3Di alphabet derived from motions observed in molecular dynamics. This original result reveals that the 3Di tokens, despite being coarse-grained descriptors of 3D structure, still offer sufficient resolution to capture aspects of conformational changes. This is evidenced by a correlation between fluctuations in the 3D protein structure over the course of a molecular dynamics trajectory and the entropy of 3Di states. Based on this insight, we introduce a proof-of-concept for making remote homology detection of proteins more sensitive by leveraging a proteins distinctive dynamic fingerprint captured by our model. Our method recovers flexibility signals with a fidelity that is biologically relevant, improving search and complementing protein structure predictions, for example, by flagging flexible, disordered, or other functionally relevant regions. Availability and ImplementationProtProfileMD is available at github.com/finnlueth/ProtProfileMD. The associated training data and model weights are available at huggingface.co/datasets/finnlueth/ProtProfileMD and huggingface.co/finnlueth/ProtProfileMD.

bioinformatics↗

bioRxiv2025

LoL-align: sensitive and fast probabilistic protein structure alignment

The ubiquitous availability of protein structures permits replacing sequence alignment with more accurate and sensitive structure alignment algorithms. LoL-align maximizes a local log-odds score for proteins to be homologous, given their intra-protein C - C distances. LoL-align is markedly more sensitive in detecting remote homologs than TMalign and DALI on single- and multi-domain benchmarks and achieves better alignment quality at 5-20x their speed. LoL-align is fully integrated into Foldseek.

bioinformatics↗

bioRxiv2025

Protein Structure Informed Bacteriophage Genome Annotation with Phold

Bacteriophage (phage) genome annotation is essential for understanding their functional potential and suitability for use as therapeutic agents. Here we introduce Phold, an annotation framework utilising protein structural information that combines the ProstT5 protein language model and structural alignment tool Foldseek. Phold assigns annotations using a database of over 1.36 million predicted phage protein structures with high quality functional labels. Benchmarking reveals that Phold outperforms existing sequence-based homology approaches in functional annotation sensitivity whilst maintaining speed, consistency and scalability. Applying Phold to diverse cultured and metagenomic phage genomes shows it consistently annotates over 50% of genes on an average phage and 40% on an average archaeal virus. Comparisons of phage protein structures to other protein structures across the tree of life reveals that phage proteins commonly have structural homology to proteins shared across the tree of life, particularly those that have nucleic acid metabolism and enzymatic functions. Phold is available as free and open-source software at https://github.com/gbouras13/phold.

bioinformatics↗

bioRxiv2025

Scaling down protein language modeling with MSA Pairformer

Recent efforts in protein language modeling have focused on scaling single-sequence models and their training data, requiring vast compute resources that limit accessibility. Although models that use multiple sequence alignments (MSA), such as MSA Transformer, offer parameter-efficient alternatives by extracting evolutionary information directly from homologous sequences rather than storing it in parameters, they generally underperform compared to single-sequence-based language due to memory inefficiencies that limit the number of sequences and averaging evolutionary signals across the MSA. We address these challenges with MSA Pairformer, a 111M parameter memory-efficient MSA-based protein language model that extracts evolutionary signals most relevant to a query sequence through bi-directional updates of sequence and pairwise representations. MSA Pairformer achieves state-of-the-art performance in unsupervised contact prediction, outperforming ESM2-15B by 6% points while using two orders of magnitude fewer parameters. In predicting contacts at protein-protein interfaces, MSA Pair-former substantially outperforms all methods with a 24% point increase over MSA Transformer. Unlike single-sequence models that deteriorate in variant effect prediction as they scale, MSA Pairformer maintains strong performance in both tasks. Ablation studies reveal triangle operations remove indirect correlations, and unlike MSA Transformer, MSA Pairformer does not hallucinate contacts after removing covariance, enabling reliable screening of interacting sequence pairs. Overall, our work presents an alternative to the current scaling paradigm in protein language modeling, enabling efficient adaptation to rapidly expanding sequence databases and opening new directions for biological discovery.

biophysics↗

bioRxiv2025

Structural motif search across the protein-universe with Folddisco

Detecting similar protein structural motifs, functionally crucial short 3D patterns, in large structure collections is computationally prohibitive. Therefore, we developed Folddisco, which overcomes this through an index of position-independent geometric features, including side-chain orientation, combined with a rarity-based scoring system. Folddiscos compact 1.45-terabyte index of 53 million AFDB50 structures enables querying this database within seconds. It is 20-fold faster in querying and 4-fold more storage-efficient than existing methods. Folddisco is more accurate and storage-efficient than state-of-the-art methods, while being an order of magnitude faster. Folddisco is free software available at folddisco.foldseek.com and a webserver at search.foldseek.com/folddisco.

bioinformatics↗

bioRxiv2025

Metagenomic-scale analysis of the predicted protein structure universe

Protein structure prediction breakthroughs, AlphaFold2 and ESMfold, have led to an unprecedented influx of computationally derived structures. The AlphaFold Protein Structure Database provides over 200 million predictions, while the ESM Metagenomic Atlas (ESMatlas) includes over 600 million predictions from uncultured microbes. We combine these into AFESM, an 820-million-entry dataset, and cluster them using a scalable pipeline based on sequence and structure similarity, yielding 5.12 million non-singleton structural clusters. We identify common ancestors and biomes for these clusters to explore their environmental diversity and specificity, and we investigate their structural novelties. From non-singleton clusters unique to ESMatlas, we identified 12 novel domain folds, repredicting a subset ([~]45%) of low-quality domains with ColabFold yielded 33 additional novel folds. This underscores the importance of prediction quality in structural novelty discovery. We also identified 11,941 previously unseen domain combinations, highlighting the untapped structural diversity and importance of metagenomic data for illuminating underexplored regions of the protein structural universe. An interactive webserver and data are available at afesm.foldseek.com.

bioinformatics↗

bioRxiv2025

Easy and interactive taxonomic profiling with Metabuli App

SummaryAccurate metagenomic taxonomic profiling is critical for understanding microbial communities. However, computational analysis often requires command-line proficiency and high-performance computing resources. To lower these barriers, we developed Metabuli App, an all-in-one desktop application that efficiently runs taxonomic profiling locally on a consumer-grade computer. It features user-friendly graphical interfaces for custom database curation, raw read quality control (QC), taxonomic profiling, and interactive result visualization. Availability and implementationGPLv3-licensed source code and prebuilt apps for Windows, macOS, and Linux are available at https://github.com/steineggerlab/Metabuli-App and are archived at https://doi.org/10.5281/zenodo.15876171. Analysis scripts are available at https://github.com/jaebeom-kim/metabuli-app-analysis. The Sankey-based taxonomy visualization component is available at https://github.com/steineggerlab/taxoview for easy integration into other web projects. Contactmartin.steinegger@snu.ac.kr

bioinformatics↗

bioRxiv2024

GPU-accelerated homology search with MMseqs2

Rapidly growing protein databases demand faster sensitive sequence similarity detection. We present GPU-accelerated search utilizing intra-query parallelization delivering 6x faster single-protein searches compared to state-of-the-art CPU methods on 2x64 cores--speeds previously requiring large protein batches. It is most cost effective, including in large-batches at 0.45x MMseqs2-CPU speed (8 GPUs delivering 2.4x). It accelerates ColabFold structure prediction 31.8x compared to AlphaFold2 and Foldseek search 4-27x. MMseqs2-GPU is open-source at mmseqs.com.

bioinformatics↗

bioRxiv2024

De novo discovery of conserved gene clusters in microbial genomes with Spacedust

Metagenomics has revolutionized environmental and human-associated microbiome studies. However, the limited fraction of proteins with known biological process and molecular functions presents a major bottleneck. In prokaryotes and viruses, evolution favors keeping genes participating in the same biological processes co-localized as conserved gene clusters. Conversely, conservation of gene neighborhood indicates functional association. Spacedust is a tool for systematic, de novo discovery of conserved gene clusters. To find homologous protein matches it uses fast and sensitive structure comparison with Foldseek. Partially conserved clusters are detected using novel clustering and order conservation P-values. We demonstrate Spacedusts sensitivity with an all-vs-all analysis of 1 308 bacterial genomes, identifying 72 843 conserved gene clusters containing 58% of the 4.2 million genes. It recovered 95% of antiviral defense system clusters annotated by a specialized tool. Spacedusts high sensitivity and speed will facilitate the annotation of the huge numbers of sequenced bacterial, archaeal and viral genomes.

bioinformatics↗

bioRxiv2024

The OMG dataset: An Open MetaGenomic corpus for mixed-modality genomic language modeling

Biological language model performance depends heavily on pretraining data quality, diversity, and size. While metagenomic datasets feature enormous biological diversity, their utilization as pretraining data has been limited due to challenges in data accessibility, quality filtering and deduplication. Here, we present the Open MetaGenomic (OMG) corpus, a genomic pretraining dataset totalling 3.1T base pairs and 3.3B protein coding sequences, obtained by combining two largest metagenomic dataset repositories (JGIs IMG and EMBLs MGnify). We first document the composition of the dataset and describe the quality filtering steps taken to remove poor quality data. We make the OMG corpus available as a mixed-modality genomic sequence dataset that represents multi-gene encoding genomic sequences with translated amino acids for protein coding sequences, and nucleic acids for intergenic sequences. We train the first mixed-modality genomic language model (gLM2) that leverages genomic context information to learn robust functional representations, as well as coevolutionary signals in protein-protein interfaces and genomic regulatory syntax. Furthermore, we show that deduplication in embedding space can be used to balance the corpus, demonstrating improved performance on downstream tasks. The OMG dataset is publicly hosted on the Hugging Face Hub at https://huggingface.co/datasets/tattabio/OMG and gLM2 is available at https://huggingface.co/tattabio/gLM2_650M.

systems biology↗

bioRxiv2024

Multiple Protein Structure Alignment at Scale with FoldMason

Protein structure is conserved beyond sequence, making multiple structural alignment (MSTA) essential for analyzing distantly related proteins. Computational prediction methods have vastly extended our repository of available proteins structures, requiring fast and accurate MSTA methods. Here, we introduce FoldMason, a progressive MSTA method that leverages the structural alphabet from Foldseek, a pairwise structural aligner, for multiple alignment of hundreds of thousands of protein structures, exceeding alignment quality of state-of-the-art methods, while two orders of magnitudes faster than other MSTA methods. FoldMason computes confidence scores, offers interactive visualizations, and provides essential speed and accuracy for large-scale protein structure analysis in the era of accurate structure prediction. Using Flaviviridae glycoproteins, we demonstrate how FoldMasons MSTAs support phylogenetic analysis below the twilight zone. FoldMason is free open-source software: foldmason.foldseek.com and webserver: search.foldseek.com/foldmason.

bioinformatics↗

bioRxiv2024

Rapid and Sensitive Protein Complex Alignment with Foldseek-Multimer

Advances in computational structure prediction will vastly augment the hundreds of thousands of currently-available protein complex structures. Translating these into discoveries requires aligning them, which is computationally prohibitive. Foldseek-Multimer computes complex alignments from compatible chain-to-chain alignments, identified by efficiently clustering their superposition vectors. Foldseek-Multimer is 3-4 orders of magnitudes faster than the gold standard, while producing comparable alignments; allowing it to compare billions of complex-pairs in 11 hours. Foldseek-Multimer is open-source software: github.com/steineggerlab/foldseek, webserver: search.foldseek.com and the BFMD database.

bioinformatics↗