About

Ramana V Davuluri is a Professor in the Department of Biomedical Informatics at Stony Brook University. His research focuses on Machine Learning applications in Cancer Data Science and Gene Regulation. He is involved in advancing computational methods to better understand biological data, particularly in the context of cancer research and genomics. His work contributes to the development of data-driven approaches for understanding complex biological systems and improving disease diagnosis and treatment.

Research topics

• Biology
• Genetics
• Cancer research
• Computational biology
• Medicine

Selected publications

• Abstract 2047: Overcoming drug resistance by novel gemcitabine-modified miR-15a in colorectal cancer.

  Cancer Research · 2026-04-03

  article

  Abstract Colorectal cancer (CRC) is the third most prevalent cancer globally. Despite therapeutic advances, chemoresistance remains a major challenge, with ∼20-30% of advanced-stage CRC patients experiencing recurrence within the first five years of treatment. Growing evidence suggests that targeting DNA damage response proteins, such as ATR/ATM, WEE1, and CHK1, is critical for overcoming resistance. In this regard, microRNAs (miRNAs) offer great therapeutic potential, as they can simultaneously target multiple signaling pathways and their dysregulation is widely implicated in cancer. Notably, tumor-suppressor miRNA-15a (miR-15a) is frequently downregulated in CRC and has been associated with poor patient prognosis. While miR-15a restoration has shown great promise, effective miRNA delivery remains a significant challenge due to its instability. Various chemical modifications have been shown to greatly enhance the stability of miRNAs. One such modification, 5-FU modified miR-15a (5-FU-miR-15a) was demonstrated as a promising therapeutic in CRC by our lab. Building on this, we engineered a new gemcitabine-modified miR-15a (Gem-miR-15a), which integrates the tumor-suppressive properties of miR-15a with the chemotherapeutic ability of gemcitabine. Though not a standard therapy in CRC, gemcitabine is emerging as an alternative for advanced refractory and resistant cases. Thus, we hypothesized that Gem-miR-15a would have an enhanced therapeutic advantage in overcoming chemoresistance to standard drugs, such as 5-FU. In this study, we assessed the effects of Gem-miR-15a on cell viability, apoptosis, and cell cycle progression of various parental (HCT116, SW480, SW620, HT-29) and 5-FU resistant CRC cell lines. Gem-miR-15a drastically reduced cell viability in both parental (IC50=1-10nM) and resistant cells (IC50=4.08 nM) without any delivery vehicle, significantly induced apoptosis, and caused an S-phase cell cycle arrest. Gem-miR-15a demonstrated dramatically increased cytotoxicity, with an almost 1000-fold reduction in IC50 as compared to the standard drug, 5-FU. The effects were consistent in 3D spheroids (IC50=9.76 nM) and patient-derived organoids (IC50=6-14 nM), too. The modification retained the target specificity of the native miR-15a, and downregulated key oncogenes like WEE1, CHK1 and BMI1 which was confirmed by western blotting. Additionally, Gem-miR-15a demonstrated strong synergy with oxaliplatin (Synergy score: 11.03). Experiments were done in triplicate (n=3) and analyzed using Student’s t-test (p < 0.05). Gem-miR-15a (4mg/kg) was also significantly able to reduce the tumor growth in in vivo metastatic mouse models, as compared to the mice treated with vehicle alone, with no visible toxicities. Hence, our findings establish Gem-miR-15a as a potent and multi-targeted therapeutic candidate capable of overcoming chemoresistance in CRC. Citation Format: Anushka Ojha, Amartya Pal, Max Chao, Ramana Davuluri, Jingfang Ju. Overcoming drug resistance by novel gemcitabine-modified miR-15a in colorectal cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 2047.

  PublisherDOI

• Solid Tumors Pan Cancer Transcriptome: Tissue/Cancer specific expression groups at the Isoform-Level

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-05-07

  articleOpen accessSenior authorCorresponding

  Most of the human genome is transcribed into diverse isoforms whose tissue specificity is profoundly disrupted in cancer, yet isoform-level dysregulation remains poorly characterized across solid tumors. Here, we introduce STPCaT (Solid Tumors Pan-Cancer Transcriptome), an isoform-centric analysis extending TransTEx to systematically classify transcript expression across TCGA solid tumors and GTEx normal tissues. STPCaT reveals a striking collapse of normal tissue-specific programs in cancer, accompanied by the emergence of two dominant expression groups: cancer-high (CanHigh) and normal-high (NorHigh) isoforms. We uncover a large repertoire of previously unannotated Cancer-Testis Antigens (CTAs), the majority of which are absent from existing CTA databases, with broad relevance across multiple cancers, including gliomas. In pan-gliomas, consensus clustering and random-forest feature selection identify compact, highly discriminative isoform signatures that robustly stratify low-grade and glioblastomas with up to 97 to 98% accuracy using as few as five transcripts. These signatures recapitulate canonical glioma biology and highlight pathways linked to migration, development, and vesicle trafficking. Independent validation in the GLASS consortium cohort demonstrates cohort-specific trends that partially recapitulate primary findings, reflecting known biological heterogeneity across patient populations. Together, STPCaT provides a scalable, isoform-resolved resource for tumor stratification, CTA discovery, and precision oncology applications across solid tumors.

  PublisherOA PDFDOI

• Development of gemcitabine-modified miR-15a as a novel, multimodal treatment strategy to overcome 5-FU and oxaliplatin resistance in colorectal cancer

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-04-29

  article

  Abstract Background Resistance to 5-fluorouracil (5-FU)-based chemotherapy is a major clinical obstacle in colorectal cancer (CRC), highlighting the urgent need to overcome established resistance mechanisms. MicroRNA-based therapeutics have emerged as compelling candidates in this context, given their inherently pleiotropic mode of action; however, their clinical translation remains hindered by poor stability and suboptimal delivery. Methods To address these limitations, Gem-miR-15a, a unique gemcitabine-modified tumor-suppressor microRNA-15a was designed to synergistically integrate the tumor-suppressive activity of miR-15a with the chemotherapeutic potency of gemcitabine into a single molecular entity. Therapeutic efficacy of Gem-miR-15a was evaluated across a spectrum of preclinical models, including parental and drug-resistant CRC cell lines, 3D tumor spheroids, patient-derived organoids and in vivo metastatic models. Cell viability, apoptosis and cell cycle analyses were performed, along with RNA sequencing and protein validation. Statistical analyses were conducted using Student’s t-test or two-way ANOVA with mixed effects, and data were presented as mean ± SD. Results Gem-miR-15a exhibited potent anti-proliferative activity with IC 50 values in the low nanomolar range, achieving ∼100-5000-fold greater potency relative to 5-FU and oxaliplatin. Importantly, it retained efficacy in both 5-FU- and oxaliplatin-resistant CRC models, effectively overcoming acquired chemoresistance. Mechanistically, Gem-miR-15a induced S-phase cell cycle arrest, eliminated the G2-phase cell population, and triggered apoptosis, accompanied by suppression of key oncogenic targets including WEE1, CHK1, YAP1 and BMI1. RNA-seq analysis further demonstrated modulation of pathways such as p53 signaling and reversal of resistance-associated gene expression, that were corroborated at the protein level. In vivo , Gem-miR-15a significantly reduced tumor growth at a dose ∼12-fold lower than gemcitabine, with no observable toxicity. Conclusion Gem-miR-15a represents a potent, multi-targeted therapeutic strategy capable of overcoming chemoresistance in CRC. Its enhanced stability, effective delivery and robust efficacy across resistant models and a favorable safety profile highlight its strong potential for clinical translation. Graphical Abstract

  PublisherDOI

• DNABERT-Enhancer Trained Models

  Zenodo (CERN European Organization for Nuclear Research) · 2026-03-22

  otherOpen accessSenior author

  This archive contains the pretrained DNABERT-Enhancer models used in the manuscript.Models included:DNABERT-Enhancer-201DNABERT-Enhancer-350These models were fine-tuned from DNABERT(https://github.com/jerryji1993/DNABERT), a large language model for the human genome, for enhancer prediction.

  PublisherDOI

• DNABERT-Enhancer Trained Models

  Zenodo (CERN European Organization for Nuclear Research) · 2026-03-22

  otherOpen accessSenior author

  This archive contains the pretrained DNABERT-Enhancer models used in the manuscript.Models included:DNABERT-Enhancer-201DNABERT-Enhancer-350These models were fine-tuned from DNABERT(https://github.com/jerryji1993/DNABERT), a large language model for the human genome, for enhancer prediction.

  PublisherDOI

• Immunoglobulin superfamily 3 (Igsf3) function is dispensable for brain development

  Scientific Reports · 2025-02-23 · 1 citations

  articleOpen access

  The Immunoglobulin superfamily (IgSF) is a heterogeneous and conserved family of adhesion proteins crucial during the development of the central nervous system including neuronal migration and synaptogenesis. The Immunoglobulin superfamily member 3 (IGSF3) is expressed in the developing brain and has been suggested to play a role during morphological development of the granule cells neurites in the cerebellum. In addition, a role for IGSF3 in supporting glioma progression has been recently demonstrated. Remaining unexplored is the physiological role of IGSF3 in regulating brain development, including neocortical development. We generated an Igsf3 knockout (KO) mouse using a CRISPR/Cas9-based approach and explored the function of Igsf3 in regulating cortical development. We found that Igsf3 largely co-localizes with other IgSF proteins during cortical development and despite its expression being developmentally regulated in neuronal progenitors and in postmitotic neurons, Igsf3 is not essential for brain development, neuronal migration, or neuronal maturation.

  PublisherOA PDFDOI

• DNABERT-S: pioneering species differentiation with species-aware DNA embeddings

  Bioinformatics · 2025-07-01 · 21 citations

  articleOpen access

  SUMMARY: We introduce DNABERT-S, a tailored genome model that develops species-aware embeddings to naturally cluster and segregate DNA sequences of different species in the embedding space. Differentiating species from genomic sequences (i.e. DNA and RNA) is vital yet challenging, since many real-world species remain uncharacterized, lacking known genomes for reference. Embedding-based methods are therefore used to differentiate species in an unsupervised manner. DNABERT-S builds upon a pre-trained genome foundation model named DNABERT-2. To encourage effective embeddings to error-prone long-read DNA sequences, we introduce Manifold Instance Mixup (MI-Mix), a contrastive objective that mixes the hidden representations of DNA sequences at randomly selected layers and trains the model to recognize and differentiate these mixed proportions at the output layer. We further enhance it with the proposed Curriculum Contrastive Learning (C2LR) strategy. Empirical results on 28 diverse datasets show DNABERT-S's effectiveness, especially in realistic label-scarce scenarios. For example, it identifies twice more species from a mixture of unlabeled genomic sequences, doubles the Adjusted Rand Index (ARI) in species clustering, and outperforms the top baseline's performance in 10-shot species classification with just a 2-shot training. AVAILABILITY AND IMPLEMENTATION: Model, codes, and data are publically available at https://github.com/MAGICS-LAB/DNABERT_S.

  PublisherOA PDFDOI

• DeepVRegulome: DNABERT-based deep-learning framework for predicting the functional impact of short genomic variants on the human regulome

  ArXiv.org · 2025-11-12

  preprintOpen accessSenior author

  Whole-genome sequencing (WGS) has revealed numerous non-coding short variants whose functional impacts remain poorly understood. Despite recent advances in deep-learning genomic approaches, accurately predicting and prioritizing clinically relevant mutations in gene regulatory regions remains a major challenge. Here we introduce Deep VRegulome, a deep-learning method for prediction and interpretation of functionally disruptive variants in the human regulome, which combines 700 DNABERT fine-tuned models, trained on vast amounts of ENCODE gene regulatory regions, with variant scoring, motif analysis, attention-based visualization, and survival analysis. We showcase its application on TCGA glioblastoma WGS dataset in prioritizing survival-associated mutations and regulatory regions. The analysis identified 572 splice-disrupting and 9,837 transcription-factor binding site altering mutations occurring in greater than 10% of glioblastoma samples. Survival analysis linked 1352 mutations and 563 disrupted regulatory regions to patient outcomes, enabling stratification via non-coding mutation signatures. All the code, fine-tuned models, and an interactive data portal are publicly available.

  PublisherOA PDFDOI

• Genomic Language Model for Predicting Enhancers and Their Allele-Specific Activity in the Human Genome

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-03-18

  preprintOpen accessSenior authorCorresponding

  Predicting and deciphering the regulatory logic of enhancers is a challenging problem, due to the intricate sequence features and lack of consistent genetic or epigenetic signatures that can accurately discriminate enhancers from other genomic regions. Recent machine-learning based methods have spotlighted the importance of extracting nucleotide composition of enhancers but failed to learn the sequence context and perform suboptimally. Motivated by advances in genomic language models, we developed DNABERT-Enhancer, a novel enhancer prediction method, by applying DNABERT pre-trained language model on the human genome. We trained two different models, using large collection of enhancers curated from the ENCODE registry of candidate cis-Regulatory Elements. The best fine-tuned model achieved 88.05% accuracy with Matthews correlation coefficient of 76% on independent set aside data. Further, we present the analysis of the predicted enhancers for all chromosomes of the human genome by comparing with the enhancer regions reported in publicly available databases. Finally, we applied DNABERT-Enhancer along with other DNABERT based regulatory genomic region prediction models to predict candidate SNPs with allele-specific enhancer and transcription factor binding activity. The genome-wide enhancer annotations and candidate loss-of-function genetic variants predicted by DNABERT-Enhancer provide valuable resources for genome interpretation in functional and clinical genomics studies.

  PublisherOA PDFDOI

• TSProm: Deciphering the Genomic Context of Tissue Specificity

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-10-31

  preprintOpen accessSenior authorCorresponding

  Abstract Characterizing tissue-specific (TSp) gene expression is crucial for understanding development and disease; however, traditional expression-based methods often overlook the latent “regulatory grammar” embedded in the non-coding DNA, particularly in distal promoter regions. Here, we introduce TSProm , a framework that specializes a DNA foundation model (DNABERT2) to decipher the long-range regulatory logic of TSp promoters at the gene isoform level. The contributions of our work are two-fold. First, we propose a novel comparative design that trains two distinct models, A: for general promoter biology and B: for TSp regulation. These models enable the precise isolation of sequence motifs around the transcription start site that uniquely define tissue identity. Second, we introduce a comprehensive explainable AI (xAI) module that integrates attention-based discovery with model-agnostic SHAP analysis to provide robust, cross-validated interpretations of learned features. Applying this framework to human brain, liver, and testis promoters, we identified and validated clinically relevant transcription factors (TFs) in the brain, including SP1, MYC , and HES6 , and confirmed their known roles in diseases such as gliomas and neuroblastomas. Our analysis further revealed that C2H2 Zinc Finger proteins are a dominant feature of the global landscape of TSp gene regulation. TSProm provides a novel and interpretable framework for identifying TSp gene regulatory elements, offering powerful computational tools for the study of tissue-specific gene regulation in normal and disease conditions.

  PublisherDOI

Recent grants

• NIH Grant R01CA135444

  NIH · $651k · 2012

• NIH Grant R01HG003362

  NIH · $1.3M · 2011

• Informatics Platform for Mammalian Gene Regulation at Isoform-level

  NIH · $2.4M · 2013–2022

• Developing Novel Deep-Learning Based Methods for Deciphering Non-Coding Gene Regulatory Code

  NIH · $1.7M · 2021–2029

Frequent coauthors

• Yingtao Bi

  AbbVie (United States)

  62 shared

• Hao Sun

  61 shared

• Sandya Liyanarachchi

  57 shared

• Milena S. Nicoloso

  Centro di Riferimento Oncologico

  56 shared

• Lianchun Xiao

  The University of Texas MD Anderson Cancer Center

  51 shared

• Christoph Plass

  Epigenomics (Germany)

  49 shared

• Carlo M. Croce

  The Ohio State University

  45 shared

• George A. Calin

  The University of Texas MD Anderson Cancer Center

  45 shared

Education

• M.S.

  Indian Agricultural Statistics Research Institute

  2001

• Ph.D.

  Indian Agricultural Statistics Research Institute

  1996