Biology subjects

Patsakis, M.

Publications and source records attributed to Patsakis, M..

10 recordsLinked to original sources

bioRxiv2026

Calibration-free compression brings Evo 2 to its full million-token context on a single GPU

Evo 2 is the largest openly available genomic foundation model, but its forty billion parameter configuration cannot be loaded onto a single 80 GB accelerator, placing genome-scale analysis beyond most laboratories. We present TurboQuant-Bio, an open toolkit that compresses Evo 2s weights and attention cache to four bits without calibration data, and serves both through fused kernels. Compression is near-lossless across perplexity spanning the tree of life, genomic classification, splice-site prediction, gene completion and clinically relevant variant-effect prediction. It brings Evo 2 40B onto one 80 GB GPU and Evo 2 7B to its full million-token context within a 40 GB memory budget, an eightfold gain in reachable context. We further show that the released chunked-prefill path is silently incorrect, returning plausible but uncorrelated likelihoods, and derive the block-wise continuation that repairs it: a complete 580-kilobase bacterial genome is now scored in one context in 22 minutes rather than 13.7 hours.

bioinformatics↗

bioRxiv2026

NYX: Format-aware, learned compression across omics file types

Genomic data repositories continue to grow as sequencing technologies improve, with the NCBI SRA alone exceeding 47 PB. General-purpose compressors treat bioinformatics files as unstructured byte streams and fail to exploit the structured nature of omics data. We present NYX, a format-aware compression system for FASTA, FASTQ, VCF, WIG, H5AD, and BED files. NYX combines lightweight, reversible preprocessing and is build upon the OpenZL framework to take advantage of inherent data structure, delivering high compression ratios while preserving fast and lossless compression. Across representative datasets in the target formats, NYX achieves substantially higher speed than format-specific compressors while maintaining or improving compression ratio.

bioinformatics↗

bioRxiv2026

Accelerating Inference in Genomic Foundation Models via Speculative Decoding

Genomic and protein foundation models (GFMs and PFMs) have demonstrated strong performance in learning the language of DNA and proteins, but their use in large-scale sequence generation is limited by the latency of autoregressive decoding. Because every token triggers a forward pass of a large Transformer, whose inference is relatively slow, long-sequence generation quickly becomes costly. In this work we adapt speculative decoding to a representative GFM: the DNA model DNAGPT and two representative PFMs: ProGen2 and ProtGPT2. We implement a probabilistic variant of speculative decoding, in which a lightweight draft model proposes short token spans and a larger target model verifies or corrects them in parallel, while preserving the target models sampling distribution. Across all three models we systematically study the effect of speculation window length, temperature, draft architecture and prompt length, and we benchmark tokens per second over multiple runs per configuration. Speculative decoding yields consistent speedups over standard key-value cached decoding, with maximum observed speedup reaching 100% increase, while average gains across models ranging between 20% and 40% (e.g. 1.2x-1.4x), without changing the underlying target model predictions. Our results show that speculative decoding is a practical and model-agnostic strategy for accelerating genomic and proteomic sequence generation without sacrificing prediction quality.

genomics↗

bioRxiv2025

Characterization of Z-DNA dynamics across the tree of life

Z-DNA/Z-RNA is an alternative left-handed nucleic acid conformation with established and emerging roles in gene regulation, immunity, and genome instability. However, its occurrence dynamics and lineage specificity across the tree of life have not yet been fully characterized. Utilizing the recently developed and improved Z-DNA searching tool, ZSeeker, we analyzed 281,139 complete organismal genomes, including multiple Telomere-to-Telomere genome assemblies, and generated genome-wide Z-nucleic acid maps, examined their topography, and compared them to dinucleotide-preserving controls. Cellular genomes featured pervasive Z-DNA enrichment relative to expectation, with enrichments of [~]1.5 and [~]1.7-fold in Bacteria and Archaea and [~]3-fold in Eukaryota. In contrast, Viruses exhibited large differences between lineages, with modest enrichment in several DNA viral groups and pronounced depletion across RNA clades, most notably Influenza A/B strains. We built a LASSO regression model trained on non-Influenza viruses (cross-validated R{superscript 2} {approx} 0.73), which identified GC content, genome type, and host type as the leading predictors for Z-nucleic acid density, yet it significantly over-predicted Z-RNA density in Influenza A/B. More than 99% of assemblies exceeded the +2 SD threshold, and a "typical Influenza" genome was predicted at 2.76 bp/kb compared to [~]0.016 bp/kb observed (a [~]170-fold overestimation based on chance alone). Together, these results reveal domain- and lineage-specific regimes: cellular genomes are enriched for Z-DNA consistent with regulatory roles, whereas influenza viruses appear to have undergone strong, lineage-specific depletion of Z-RNA-forming sequences, likely reflecting evolutionary pressure tied to host sensing pathways.

evolutionary biology↗

bioRxiv2025

Quantum implementation of multi-pattern string matching for k-mer detection

MotivationThe exponential growth of publicly available genomic data has created unprecedented opportunities for sequence-based discovery. Locating specific k-mers is fundamental to diverse applications, including metagenomic classification, pathogen and cancer detection, and variant calling yet efficient identification of multiple k-mer patterns across large sequencing data and massive databases remains a significant computational challenge. MethodWe implement two quantum algorithms for DNA multi pattern string matching for k-mer detection based on Grovers amplitude amplification with quantum random access memory (QRAM). The first algorithm uses an enumerate-m oracle that sequentially checks a loaded text substring against all m patterns achieving O({surd}S) query complexity for S text positions but requiring O(m{middle dot}L) work per oracle call. The second algorithm employs nested Grover search with an outer loop over text positions and an inner loop over pattern space, reducing oracle complexity to O(L) while performing O({surd}S {middle dot} {surd}m) in total. ResultsWe present two quantum implementations of multi-pattern string matching tailored for k-mer detection. Leveraging quantum parallelism and Grover-inspired search primitives, our methods accelerate dictionary-based pattern matching, particularly in contexts involving large sequences, such as genomic data, and extensive pattern sets. ConclusionsWhile implementation challenges such as QRAM overhead remain, this study demonstrates both the promise and current limitations of quantum-enhanced string matching, establishing a foundational step toward quantum readiness in bioinformatics. Availability and implementationTo maximize accessibility and practical use, we provide our methodology at: https://github.com/Georgakopoulos-Soares-lab/quantum-multi-motif-finder

bioinformatics↗

bioRxiv2025

Unraveling diversity by isolating peptide sequences specific to distinct taxonomic groups

The identification of succinct, universal fingerprints that enable the characterization of individual taxonomies can reveal insights into trait development and can have widespread applications in pathogen diagnostics, human healthcare, ecology and the characterization of biomes. Here, we investigated the existence of peptide k-mer sequences that are exclusively present in a specific taxonomy and absent in every other taxonomic level, termed taxonomic quasi-primes. By analyzing proteomes across 24,073 species, we identified quasi-prime peptides specific to superkingdoms, kingdoms, and phyla, uncovering their taxonomic distributions and functional relevance. These peptides exhibit remarkable sequence uniqueness at six- and seven-amino- acid lengths, offering insights into evolutionary divergence and lineage-specific adaptations. Moreover, we show that human quasi-prime loci are more prone to harboring pathogenic variants, underscoring their functional significance. This study introduces taxonomic quasi-primes and offers insights into their contributions to proteomic diversity, evolutionary pathways, and functional adaptations across the tree of life, while emphasizing their potential impact on human health and disease.

bioinformatics↗

bioRxiv2024

invertiaDB: A Database of Inverted Repeats Across Organismal Genomes

Inverted repeats are repetitive elements that can form hairpin and cruciform structures. They are linked to genomic instability, however they also have various biological functions. Their distribution differs markedly across taxonomic groups in the tree of life, and they exhibit high polymorphism due to their inherent genomic instability. Advances in sequencing technologies and declined costs have enabled the generation of an ever-growing number of complete genomes for organisms across taxonomic groups in the tree of life. However, a comprehensive database encompassing inverted repeats across diverse organismal genomes has been lacking. We present InvertiaDB, the first comprehensive database of inverted repeats spanning multiple taxa, featuring repeats identified in the genomes of 118,070 organisms across all major taxonomic groups. The database currently hosts 30,067,666 inverted repeat sequences, serving as a centralized, user-friendly repository to perform searches, interactive visualization, and download existing inverted repeat data for independent analysis. invertiaDB is implemented as a web portal for browsing, analyzing and downloading inverted repeat data. invertiaDB is publicly available at https://invertiadb.netlify.app/homepage.html.

genomics↗

bioRxiv2024

Ribosomal DNA arrays are the most H-DNA rich element in the human genome

Repetitive DNA sequences can form non-canonical structures such as H-DNA which is an intramolecular triplex DNA structure. The new Telomere-to-Telomere (T2T) genome assembly for the human genome has eliminated gaps, enabling the examination of highly repetitive regions including centromeric and pericentromeric repeats and ribosomal DNA arrays. This gapless assembly allows for the examination of the distribution of H-DNA sequences in parts of the human genome that were not previously annotated. We find that H-DNA appears once every 30,000 bps in the human genome. Its distribution is highly inhomogeneous with H-DNA motif hotspots being detectable in acrocentric chromosomes. Ribosomal DNA arrays in acrocentric chromosomes are the genomic element with the highest H-DNA enrichment, with 13.22% of total H-DNA motifs being found in ribosomal DNA arrays, representing a 42.65-fold enrichment over what would be expected by chance. Across the acrocentric chromosomes we report that 55.87% of all H-DNA motifs found in these chromosomes are in rDNA array loci. The H-DNA motifs are primarily found in the intergenic spacer regions of the ribosomal DNA arrays, generating repeated clusters. We also discover that binding sites for PRDM9, a protein that regulates the formation of double-strand breaks and determines the meiotic recombination hotspots in humans and most mammals, are over 5-fold enriched for H-DNA motifs. Finally, we provide evidence that our findings are consistent in other non-human great ape genomes. We conclude that ribosomal DNA arrays are the most enriched genomic loci for H-DNA sequences in human and other great ape genomes.

genomics↗

bioRxiv2024

Quadrupia: Derivation of G-quadruplexes for organismal genomes across the tree of life

G-quadruplex DNA structures exhibit a profound influence on essential biological processes, including transcription, replication, telomere maintenance, and genomic stability. These structures have demonstrably shaped organismal evolution. However, a comprehensive, organism-wide G-quadruplex map encompassing the diversity of life has remained elusive. Here, we introduce Quadrupia, the most extensive and well-characterized G-quadruplex database to date, facilitating the exploration of G-quadruplex structures across the evolutionary spectrum. Quadrupia has identified G-quadruplex sequences in 108,449 reference genomes, with a total of 140,181,277 G-quadruplexes. The database also hosts a collection of 319,784 G-quadruplex clusters of 20 or more members, annotated by taxonomic distributions, multiple sequence alignments, profile Hidden Markov Models and cross-references to G-quadruplex 3D structures. Examination of G-quadruplexes across functional genomic elements in different taxa indicates preferential orientation and positioning, with significant differences between individual taxonomic groups. For example, we find that G-quadruplexes in bacteria with a single replication origin display profound preference for the leading orientation. Finally, we experimentally validate the most frequently observed G-quadruplexes using CD-spectroscopy, UV melting, and fluorescent-based approaches. Quadrupia is publicly available through https://www.pavlopoulos-lab.org/quadrupia.

genomics↗

bioRxiv2023

Nucleic Quasi-Primes: Identification of the Shortest Unique Oligonucleotide Sequences in a Species

Despite the exponential increase in sequencing information driven by massively parallel DNA sequencing technologies, universal and succinct genomic fingerprints for each organism are still missing. Identifying the shortest species-specific nucleic sequences offers insights into species evolution and holds potential practical applications in agriculture, wildlife conservation, and healthcare. We propose a new method for sequence analysis termed nucleic "quasi-primes", the shortest occurring sequences in each of 45,785 organismal reference genomes, present in one genome and absent from every other examined genome. In the human genome, we find that the genomic loci of nucleic quasi-primes are most enriched for genes associated with brain development and cognitive function. In a single-cell case study focusing on the human primary motor cortex, nucleic quasi-prime genes account for a significantly larger proportion of the variation based on average gene expression. Non-neuronal cell-types, including astrocytes, endothelial cells, microglia perivascular-macrophages, oligodendrocytes, and vascular and leptomeningeal cells, exhibited significant activation of quasi-prime containing gene associations related to cancer, while simultaneously suppressing quasi-prime containing genes were associated with cognitive, mental, and developmental disorders. We also show that human disease-causing variants, eQTLs, mQTLs and sQTLs are 4.43-fold, 4.34-fold, 4.29-fold and 4.21-fold enriched at human quasi-prime loci, respectively. These findings indicate that nucleic quasi-primes are genomic loci linked to the evolution of species-specific traits and in humans they provide insights in the development of cognitive traits and human diseases, including neurodevelopmental disorders.

genomics↗