Biology subjects

Alipanahi, B.

Publications and source records attributed to Alipanahi, B..

10 recordsLinked to original sources

bioRxiv2025

Orphan Non-Coding RNAs Drive Tumorigenesis and Enable Pre-Diagnostic Detection in HPV-Negative Head and Neck Cancer

HPV-negative head and neck squamous cell carcinoma (HNSCC) lacks reliable biomarkers for early detection and molecular stratification. We recently described orphan non-coding RNAs (oncRNAs) as a class of cancer-emergent small RNAs that arise from cryptic promoters largely silent in normal tissues. Here, we define the comprehensive landscape of oncRNAs in HNSCC. Using TCGA and CPTAC-3 data, we demonstrate that these transcripts are not only tumor-specific but provide digital fingerprints that accurately stratify tumors by anatomical site. Moving from association to function, we investigated whether these emergent transcripts actively drive tumorigenesis. We focused on oncRNA64585, a transcript highly enriched in oral cavity tumors and associated with tumor progression. Its depletion significantly impaired proliferation in vitro and suppressed tumor growth in syngeneic murine models. Mechanistically, transcriptomic and chromatin accessibility profiling revealed that oncRNA64585 regulates transcriptional programs governing cell cycle progression and apoptosis through interaction with the Polycomb Repressive Complex 2 (PRC2). Finally, we demonstrate the clinical utility of these signatures: oncRNA profiles in plasma predicted oral cavity cancer with >90% accuracy up to four years prior to clinical diagnosis. These findings establish oncRNAs as active drivers of HNSCC biology and potent tools for early detection and therapeutic targeting.

cancer biology↗

bioRxiv2025

A multi-modal cell-free RNA language model for liquid biopsy applications

Cell-free RNA (cfRNA) profiling has emerged as a powerful tool for non-invasive disease detection, but its application is limited by data sparsity and complexity, especially in settings with constrained sample availability. We introduce Exai-1, a multi-modal, transformer-based generative foundation model that integrates RNA sequence embeddings with cfRNA abundance data to capture biologically meaningful representations of circulating RNAs. By leveraging both sequence and expression modalities, Exai-1 captures a biologically meaningful latent structure of cfRNA profiles. Pre-trained on over 306 billion tokens from 8,339 samples, Exai-1 enhances signal fidelity, reduces technical noise, and improves disease detection by generating synthetic cfRNA profiles. We show that self-attention and variational inference are particularly important for preservation of biological signals and contextual relationships. Additionally, Exai-1 facilitates cross-biofluid translation and assay compatibility through disentangling biological signals from confounders. By uniting sequence-informed embeddings with cfRNA expression patterns, Exai-1 establishes a transfer learning foundation for liquid biopsy, offering a scalable and adaptable framework for next-generation cfRNA-based diagnostics.

bioinformatics↗

bioRxiv2023

High-throughput ML-guided design of diverse single-domain antibodies against SARS-CoV-2

Treating rapidly evolving pathogenic diseases such as COVID-19 requires a therapeutic approach that accommodates the emergence of viral variants over time. Our machine learning (ML)-guided sequence design platform combines high-throughput experiments with ML to generate highly diverse single-domain antibodies (VHHs) that bind and neutralize SARS-CoV-1 and SARS-CoV-2. Crucially, the model, trained using binding data against early SARS-CoV variants, accurately captures the relationship between VHH sequence and binding activity across a broad swathe of sequence space. We discover ML-designed VHHs that exhibit considerable cross-reactivity and successfully neutralize targets not seen during training, including the Delta and Omicron BA.1 variants of SARS-CoV-2. Our ML-designed VHHs include thousands of variants 4-15 mutations from the parent sequence with significantly improved activity, demonstrating that ML-guided sequence design can successfully navigate vast regions of sequence space to unlock and future-proof potential therapeutics against rapidly evolving pathogens.

bioinformatics↗

bioRxiv2021

DeepNull: Modeling non-linear covariate effects improves phenotype prediction and association power

Genome-wide association studies (GWAS) examine the association between genotype and phenotype while adjusting for a set of covariates. Although the covariates may have non-linear or interactive effects, due to the challenge of specifying the model, GWAS often neglect such terms. Here we introduce DeepNull, a method that identifies and adjusts for non-linear and interactive covariate effects using a deep neural network. In analyses of simulated and real data, we demonstrate that DeepNull maintains tight control of the type I error while increasing statistical power by up to 20% in the presence of non-linear and interactive effects. Moreover, in the absence of such effects, DeepNull incurs no loss of power. When applied to 10 phenotypes from the UK Biobank (n=370K), DeepNull discovered more hits (+6%) and loci (+7%), on average, than conventional association analyses, many of which are biologically plausible or have previously been reported. Finally, DeepNull improves upon linear modeling for phenotypic prediction (+23% on average).

genetics↗

bioRxiv2021

Buffering Updates Enables Efficient Dynamic de Bruijn Graphs

MotivationThe de Bruijn graph has become a ubiquitous graph model for biological data ever since its initial introduction in the late 1990s. It has been used for a variety of purposes including genome assembly (Zerbino and Birney, 2008; Bankevich et al., 2012; Peng et al., 2012), variant detection (Alipanahi et al., 2020b; Iqbal et al., 2012), and storage of assembled genomes (Chikhi et al., 2016). For this reason, there have been over a dozen methods for building and representing the de Bruijn graph and its variants in a space and time efficient manner. ResultsWith the exception of a few data structures (Muggli et al., 2019; Holley and Melsted, 2020; Crawford et al., 2018), compressed and compact de Bruijn graphs do not allow for the graph to be efficiently updated, meaning that data can be be added or deleted. The most recent compressed dynamic de Bruijn graph (Alipanahi et al., 2020a), relies on dynamic bit vectors which are slow in theory and practice. To address this shortcoming, we present a compressed dynamic de Bruijn graph that removes the necessity of dynamic bit vectors by buffering data that should be added or removed from the graph. We implement our method, which we refer to as BufBOSS, and compare its performance to Bifrost, DynamicBOSS, and FDBG. Our experiments demonstrate that BufBOSS achieves attractive trade-offs compared to other tools in terms of time, memory and disk, and has the best deletion performance by an order of magnitude. Contactalanko.jarno@gmail.com

bioinformatics↗

bioRxiv2020

Succinct Dynamic de Bruijn Graphs

MotivationThe de Bruijn graph is one of the fundamental data structures for analysis of high throughput sequencing data. In order to be applicable to population-scale studies, it is essential to build and store the graph in a space- and time-efficient manner. In addition, due to the ever-changing nature of population studies, it has become essential to update the graph after construction e.g. add and remove nodes and edges. Although there has been substantial effort on making the construction and storage of the graph efficient, there is a limited amount of work in building the graph in an efficient and mutable manner. Hence, most space efficient data structures require complete reconstruction of the graph in order to add or remove edges or nodes. ResultsIn this paper we present DynamicBOSS, a succinct representation of the de Bruijn graph that allows for an unlimited number of additions and deletions of nodes and edges. We compare our method with other competing methods and demonstrate that DynamicBOSS is the only method that supports both addition and deletion and is applicable to very large samples (e.g. greater than 15 billion k-mers). Competing dynamic methods e.g., FDBG (Crawford et al., 2018) cannot be constructed on large scale datasets, or cannot support both addition and deletion e.g., BiFrost (Holley and Melsted, 2019). AvailabilityDynamicBOSS is publicly available at https://github.com/baharpan/dynboss. Contactbaharpan@ufl.edu

bioinformatics↗

bioRxiv2019

Human loss-of-function variants suggest that partial LRRK2 inhibition is a safe therapeutic strategy for Parkinsons disease

Human genetic variants causing loss-of-function (LoF) of protein-coding genes provide natural in vivo models of gene inactivation, which are powerful indicators of gene function and the potential toxicity of therapeutic inhibitors targeting these genes1,2. Gain-of-kinase-function variants in LRRK2 are known to significantly increase the risk of Parkinsons disease3,4, suggesting that inhibition of LRRK2 kinase activity is a promising therapeutic strategy. Whilst preclinical studies in model organisms have raised some on-target toxicity concerns5-8, the biological consequences of LRRK2 inhibition have not been well characterized in humans. Here we systematically analyse LoF variants in LRRK2 observed across 141,456 individuals sequenced in the Genome Aggregation Database (gnomAD)9 and over 4 million participants in the 23andMe genotyped dataset, to assess their impact at a molecular and phenotypic level. After thorough variant curation, we identify 1,358 individuals with high-confidence predicted LoF variants in LRRK2, several with experimental validation. We show that heterozygous LoF of LRRK2 reduces LRRK2 protein level by ~50% but is not associated with reduced life expectancy, or with any specific phenotype or disease state. These data suggest that therapeutics that downregulate LRRK2 levels or kinase activity by up to 50% are unlikely to have major on-target safety liabilities. Our results demonstrate the value of large scale genomic databases and phenotyping of human LoF carriers for target validation in drug discovery.

genomics↗

bioRxiv2018

The Parkinson’s Phenome: Traits Associated with Parkinson’s Disease in a Large and Deeply Phenotyped Cohort

BackgroundObservational studies have begun to characterize the wide spectrum of phenotypes associated with Parkinsons disease (PD), but recruiting large numbers of PD cases and assaying a diversity of phenotypes has often been difficult. Here, we set out to systematically describe the PD phenome using a cross-sectional case-control design in a large database.\n\nMethodsWe analyzed the association between PD and 840 phenotypes derived from online surveys. For each phenotype, we ran a logistic regression using an average of 5,141 PD cases and 65,459 age- and sex-matched controls. We selected uncorrelated phenotypes, determined statistical significance after correcting for multiple testing, and systematically assessed the novelty of each significant association. We tested whether significant phenotypes were also associated with disease duration in PD cases.\n\nFindingsPD diagnosis was associated with 149 independent phenotypes. We replicated 32 known associations and discovered 49 associations that have not previously been reported. We found that migraine, obsessive-compulsive disorder, seasonal allergies, and anemia were associated with PD, but were not significantly associated with PD duration, and tend to occur decades before the average age of diagnosis for PD. Further work is needed to determine whether these phenotypes are PD risk factors or whether they share common disease mechanisms.\n\nInterpretationWe used a systematic approach in a single large dataset to assess the spectrum of traits that were associated with PD. Some of these traits may be risk factors for PD, features of the pre-diagnostic phase of disease, or manifestations of PD pathology. The model outputs from all 840 logistic regressions are available to the research community and may be used to generate hypotheses regarding PD etiology.\n\nFundingThe Michael J. Fox Foundation, Parkinsons UK, Barts Charity, National Institute on Aging, and 23andMe, Inc.\n\nResearch in ContextO_ST_ABSEvidence before this studyC_ST_ABSWe used PubMed to perform a MEDLINE database search for review articles published up to January 21st, 2018 that contained the keywords \"Parkinson\" and \"epidemiology\" in the title or abstract. We performed additional MEDLINE searches for each phenotype that was significantly associated with PD. Although dozens of phenotypes have been tested for an association with PD, only a few associations have been consistently repeatable (e.g. pesticide exposure, coffee consumption).\n\nAdded value of this studyWe systematically tested for an association between PD and 840 phenotypes using up to 13,546 cases and 1{middle dot}3 million controls, making this one of the largest PD epidemiology studies ever conducted. We discovered 49 novel associations that will need to be replicated or validated. We found 44 associations for phenotypes that have previously been studied in relation to PD, but for which an association has not been consistently demonstrated.\n\nImplications of all the available evidenceTaken together with results from previous studies, this series of case-control analyses adds evidence for associations between PD and many phenotypes that are not currently thought to be part of the canonical PD phenome. This work paves the way for future studies to assess whether any of these phenotypes represent PD risk factors and whether any of these risk factors are modifiable.

epidemiology↗

bioRxiv2017

Resistome SNP Calling via Read Colored de Bruijn Graphs

MotivationThe resistome, which refers to all of the antimicrobial resistance (AMR) genes in pathogenic and non-pathogenic bacteria, is frequently studied using shotgun metagenomic data [14, 47]. Unfortunately, few existing methods are able to identify single nucleotide polymorphisms (SNPs) within metagenomic data, and to the best of our knowledge, no methods exist to detect SNPs within AMR genes within the resistome. The ability to identify SNPs in AMR genes across the resistome would represent a significant advance in understanding the dissemination and evolution of AMR, as SNP identification would enable \"fingerprinting\" of the resistome, which could then be used to track AMR dynamics across various settings and/or time periods.\n\nResultsWe present LueVari, a reference-free SNP caller based on the read colored de Bruijn graph, an extension of the traditional de Bruijn graph that allows repeated regions longer than the k-mer length and shorter than the read length to be identified unambiguously. We demonstrate LueVari was the only method that had reliable sensitivity (between 73% and 98%) as the performance of competing methods varied widely. Furthermore, we show LueVari constructs sequences containing the variation which span 93% of the gene in datasets with lower coverage (15X), and 100% of the gene in datasets with higher coverage (30X).\n\nAvailabilityCode and datasets are publicly available at https://github.com/baharpan/cosmo/tree/LueVari.

bioinformatics↗

bioRxiv2016

Disentangled Long-Read De Bruijn Graphs via Optical Maps

Pacific Biosciences (PacBio), the main third generation sequencing technology can produce scalable, high-throughput, unprecedented sequencing results through long reads with uniform coverage. Although these long reads have been shown to increase the quality of draft genomes in repetitive regions, fundamental computational challenges remain in overcoming their high error rate and assembling them efficiently. In this paper we show that the de Bruijn graph built on the long reads can be efficiently and substantially disentangled using optical mapping data as auxiliary information. Fundamental to our approach is the use of the positional de Bruijn graph and a succinct data structure for constructing and traversing this graph. Our experimental results show that over 97.7% of directed cycles have been removed from the resulting positional de Bruijn graph as compared to its non-positional counterpart. Our results thus indicate that disentangling the de Bruijn graph using positional information is a promising direction for developing a simple and efficient assembly algorithm for long reads.

bioinformatics↗