bioRxiv Science⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.64898/2026.06.03.729798

DeltaMut: An Integrative Database of AlphaFold2-Derived Missense Variant Structures

Qorri, E.·Adam, K.·Takacs, B.·Shemesh, S.·Buzafalvi, D.·Varga, V.·Pekker, E.·Pinter, L.·Hegedüs, Z.·Csanyi, B.·Haracska, L.
View original source ↗Export metadata ↓

Abstract

The widespread use of next-generation sequencing has led to a surge in the number of identified variants with uncertain effects on protein function. These variants pose a significant challenge in diagnostics and hinder patient treatment strategies. Numerous variant effect predictors (VEPs) are available to assess variant impact, but they primarily rely on sequence-derived information. The recent development of AlphaFold2 has raised questions about whether information retrieved from wild-type or predicted structures of missense variants can improve the predictive power of these algorithms. While the AlphaFold Protein Structure Database serves as a valuable resource for wild-type protein structures, a large-scale collection of missense variant structures is not available, limiting current efforts to wild-type conformations and a handful of modeled variants. To address this limitation, we developed DeltaMut, a comprehensive database containing over 77,000 protein structures, including 65,000 pathogenic and neutral missense variants. All structural models were generated using ParaFold, a high-performance computing-optimized implementation of AlphaFold2. The large-scale and systematic generation of variant protein structures distinguish DeltaMut as a unique resource for both expansive statistical studies and detailed, case-specific investigations of variant-induced structural changes. Furthermore, the DeltaMut database is freely accessible without registration. HighlightsO_LIDeltaMut is currently the largest database of AlphaFold2-predicted variant structures. C_LIO_LIContains 77,713 structures covering 12,101 wild-type and 65,612 variant proteins. C_LIO_LI70.6% of predicted structures have high or very high confidence (pLDDT [≥] 70). C_LIO_LIFreely accessible web server with visualization and download of variant models. C_LI

Explore related subjects

bioinformaticsPreprint

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Qorri, E., Adam, K., Takacs, B., Shemesh, S., Buzafalvi, D., Varga, V., Pekker, E., Pinter, L., Hegedüs, Z., Csanyi, B., Haracska, L.. 2026-06-04. DeltaMut: An Integrative Database of AlphaFold2-Derived Missense Variant Structures. https://doi.org/10.64898/2026.06.03.729798

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

CellChat Hotspot: A Targeted Approach to Cell Cell Communication Analysis With Spatial transcriptomics Validation.

Abstract. In recent years, single cell RNA sequencing (scRNA seq) has enabled the investigation of intercellular communication through Cell Cell Communication (CCC) tools such as CellChat. Despite their potential, these approaches remain challenging to interpret, particularly when global analyses generate complex interaction networks that may include biologically irrelevant signals. To address this limitation, we developed a framework termed CellChat Hotspot, designed to improve the biological interpretability of inferred communication networks. This approach integrates gene set enrichment analysis with a context driven filtering strategy, selecting both genes and cells based on their functional relevance prior to communication inference. Applied to stage II colon cancer data, CellChat Hotspot enabled the identification of condition-specific interaction patterns associated with recurrence. Non recurrent patients showed enrichment of immune related processes, whereas recurrent cases were characterized by pathways associated with invasion and metastasis. Notably, a loss of interaction in the HSPG pathway between CALB1+ and stromal cells was observed in recurrent patients, suggesting a potential disruption of cell matrix adhesion mechanisms. To verify the reliability of this finding, we applied the framework to spatial transcriptomics data, which confirmed the presence of the HSPG2 DAG1 interaction in non-recurrent tumors and its near absence in recurrent ones. Overall, this framework provides a biologically guided strategy for studying intercellular communication in complex systems such as the tumor microenvironment.

bioinformatics↗
bioRxiv2026

SIFT: Syncmer-Informed Selection of Haplotypes for T2T Imputation

Genome patching improves the completeness of fragmented assemblies by filling unresolved gaps with sequences from reference genomes. Pangenome panels provide diverse candidate haplotypes for this task, but identifying suitable references efficiently becomes increasingly challenging as these panels grow. We present SIFT, a compressed indexing method for selecting reference haplotypes for genome patching with ImpuT2T. SIFT combines closed-syncmer parsing, phrase-level indexing, and pseudo-maximal exact match (pseudo-MEM) queries. Closed syncmers provide a context-independent parsing rule with bounded distances between consecutive selected positions, enabling consistent parsing of reference and query sequences. We demonstrate that two closed syncmers determine the distance at which they can occur consecutively in any string, so that equal substrings of two parses represent identical sequences and no gap sequence need be stored. SIFT indexes the reference parse and its reverse using run-length compressed suffix arrays and adapts the Boyer-Moore-Li search procedure to identify pseudo-MEMs between the reference parse and query parse. It then aggregates matching evidence to identify candidate reference haplotypes for downstream patching. This framework connects compressed sequence indexing with haplotype selection and provides a basis for querying large reference panels at the phrase level. SIFT is publicly available under the GNU license at https://github.com/EddieFerro/SIFT.

bioinformatics↗
bioRxiv2026

Assessing the Capacity of Genomic Language Models to Generate Informative Embeddings of Viral Genome Fragments

Genomic language models (gLMs) trained on large volumes of nucleotide sequences are becoming an alternative to classical methods in metagenomic analysis, particularly for studying genomic "dark matter". In this work, we investigate the ability of four gLMs (DNABERT-2, DNABERT-S, ViraLM, and Evo2-7B) to generate taxonomically informative vector representations of short viral genome fragments. We constructed a balanced reference dataset of viruses based on the ICTV Virus Metadata Resource and compared the representations generated by gLMs with conventional k-mer profiles and MinHash sketches using clustering and neighborhood-based metrics (adjusted Rand index (ARI), precision@k, and CROSS precision@k). On the reference dataset, three of the four gLMs outperformed the 4-mer profile in terms of ARI and precision@5 under the same evaluation protocol. The best performance was generally achieved by intermediate model layers. Conventional k-mer profiles consistently outperformed MinHash sketches. We further evaluated the methods on two viral orders (egoviruses and mriyaviruses) absent from the reference dataset, with model layers and k-mer lengths selected exclusively on the reference dataset. DNABERT-S showed the most consistently strong performance across these two orders, ranking first on mriyaviruses and second on egoviruses, while k-mer profiles with Jensen-Shannon distance remained competitive. We also observed the consistency of representations between forward fragments and their reverse complements. Overall, the best-performing gLMs surpassed conventional baselines, although their advantage depended on the dataset and evaluation metric. The consistently strong performance of DNABERT-S suggests a potential benefit of contrastive fine-tuning and indicates that larger models are not necessarily more effective for generating taxonomically informative representations.

bioinformatics↗