About

Ludmil B. Alexandrov is a Professor in the Department of Cellular and Molecular Medicine and the Department of Bioengineering at the University of California San Diego. His research is focused on disentangling the enigmatic secrets hidden in large omics datasets by developing novel machine-learning approaches. He leverages these approaches to elucidate the basic molecular mechanisms underlying cancer development and progression, with the aim of improving cancer treatment and prevention.

Research topics

• Biology
• Genetics
• Computational biology
• Evolutionary biology
• Mathematics
• Internal medicine
• Cancer research
• Oncology
• Statistics
• Virology
• Medicine

Selected publications

• Abstract 2202: APOBEC3C rewires RNA splicing and self-renewal in hematopoietic stem and progenitor cells.

  Cancer Research · 2026-04-03

  article

  Abstract IntroductionInflammatory cytokine-responsive APOBEC3 cytosine deaminases promote antiviral defense through DNA and RNA editing and, when deregulated, drive somatic mutations and cancer progression. While APOBEC3A and APOBEC3B are well studied, the role of APOBEC3C remains less defined. However, our recent whole genome and whole transcriptome analyses of myeloproliferative neoplasm (MPN) derived hematopoietic stem and progenitor cells (HSPCs) revealed that APOBEC3C is highly expressed in this context, suggesting a context-specific function in hematopoiesis. This observation prompted us to investigate the effects of APOBEC3C, alongside other APOBEC family members, on HSPC biology, with an emphasis on their roles as drivers of clonal hematopoiesis (CH) and myeloid disorders in the context of aging and inflammation. MethodsWe collected healthy cord blood (CB) and aged bone marrow (ABM), and lentivirally transduced immunomagnetic bead-selected CD34+ cells with APOBEC3B, C, D, F, or G, or pCDH lentiviral backbone controls. Subsequently, we performed whole genome and transcriptome analyses and assessed C-to-T DNA mutations, C-to-U RNA edits and differential gene expression. As widespread changes in RNA splicing where observed as well as widespread changes in expression of genes implicated in CH upon lentiviral transduction of APOBEC3C, we also performed dual fluorescent splicing reporter assays, survival and self-renewal assays, and in vivo engraftment studies, respectively. ResultsWe discovered that APOBEC3C and 3F, induce more DNA mutations, whereas APOBEC3B, 3C and 3G disrupt RNA splicing. We observed differential expression of genes crucial for hematopoiesis, such as ZRSR2, U2AF1, and SRSF2, as well as differential exon usage and enrichment of the spliceosome pathway. Additionally, we demonstrated that APOBEC3C induced substantially greater transcriptomic effects in ABM HSPCs compared to CB HSPCs. We observed a 10-fold increase in differential gene expression, including alterations in ribosomal genes (RPS19 and RPL5), alongside splicing factors (U2AF1, ZRSR2, and SF3B2) implicated in CH, MPN, and myelodysplastic syndrome (MDS) pathogenesis. Moreover, in ABM HSPCs, APOBEC3C was found to affect CH-associated transcripts, including DNMT3A, enhance ADAR1-mediated RNA editing, and increase self-renewal capacity. Consistent with these findings, APOBEC3C is elevated in JAK2+ MPN HSPCs, while its knockdown reduced JAK2+ MPN CD34+ self-renewal and ADAR1 reporter activity. ConclusionIn conclusion, these findings demonstrate that APOBEC3C modulates RNA splicing and C-to-U RNA editing in CB derived HSPCs, with more pronounced effects in ABM derived HSPCs. In ABM, this also includes increased ADAR1p150 expression and elevated A-to-I editing, which combined with increased HSPC expansion, may contribute to the pathogenesis and progression of myeloid disorders. Citation Format: Inge van der Werf, Jane Isquith, Emma Klacking, Jessica Pham, Wenxue Ma, Shuvro P. Nandi, Rongjie Wu, Claire Engstrom, Neha Katragadda, Anna A. Khachatrian, Thomas Whisenant, Ludmil Alexandrov, Catriona Jamieson. APOBEC3C rewires RNA splicing and self-renewal in hematopoietic stem and progenitor cells [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 2202.

  PublisherDOI

• CarD-T: LLM Automated Literature Review for the Nomination & Analysis of Potential Human Carcinogens

  Zenodo (CERN European Organization for Nuclear Research) · 2026-04-07

  articleOpen access

  Carcinogenic Determination via Transformers (or CarD-T) is an automated pipeline that combines transformer-based machine learning with probabilistic analysis to identify potential carcinogens from biomedical literature. The framework processes accumulating scientific publications (left), applies a trained Named Entity Recognition (NER) model to extract potential carcinogenic entities (center), and Probabilistic Carcinogen Denomination (or PCarD) to analyze temporal trends in evidence shifts (right). This approach enables classification of candidates through Bayesian temporal analysis, overcoming limitations of traditional manual literature review methods.

  PublisherDOI

• CarD-T: LLM Automated Literature Review for the Nomination & Analysis of Potential Human Carcinogens

  Zenodo (CERN European Organization for Nuclear Research) · 2026-04-13

  articleOpen access

  Carcinogenic Determination via Transformers (or CarD-T) is an automated pipeline that combines transformer-based machine learning with probabilistic analysis to identify potential carcinogens from biomedical literature. The framework processes accumulating scientific publications (left), applies a trained Named Entity Recognition (NER) model to extract potential carcinogenic entities (center), and Probabilistic Carcinogen Denomination (or PCarD) to analyze temporal trends in evidence shifts (right). This approach enables classification of candidates through Bayesian temporal analysis, overcoming limitations of traditional manual literature review methods.

  PublisherDOI

• Systemic mutagen exposures reported by normal kidney cell genomes

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-04-09

  articleOpen access

  Abstract Lifestyle, environmental and other exposures to exogenous mutagens generate somatic mutations in normal human cells in vivo and increase cancer risk. However, the global repertoire of exogenous mutagen exposures is uncertain. The mutational signatures of mutagens in normal tissues offer opportunities to detect such exposures and survey them at population level. Using single-molecule duplex sequencing of normal kidney (n=319) and blood (n=272) samples from 10 countries, we show that normal kidney cell genomes report an extensive repertoire of somatic mutational signatures. Microdissection of kidney structures revealed that proximal tubules exhibit higher mutation rates than other components of the nephron and most normal cell types despite low cell division rates. This is explained by marked enrichment of mutational signatures due to known exogenous carcinogenic mutagens including the plant-derived aristolochic acids, as well as several signatures of unknown causes including an unknown agent prevalent in Japan (SBS12), and signatures of uncertain origins (SBS40b and SBS40c). The results suggest the existence of multiple, common, systemically circulating mutagens affecting human populations and indicate that the genomes of kidney proximal tubule cells report such exposures with high sensitivity.

  PublisherDOI

• Abstract 1387: Live cell visualization of RNA splicing dynamics upon APOBEC3 lentiviral transduction in aged bone marrow using a 3D biosensing nanobioreactor.

  Cancer Research · 2026-04-03

  article

  Abstract Background: During aging, hematopoietic stem and progenitor cells (HSPCs) undergo a progressive decline in regenerative capacity, exhibit skewed differentiation toward the myeloid lineage, show increased sensitivity to stressors, and accumulate somatic mutations, a process known as clonal hematopoiesis (CH), which increases susceptibility to hematological malignancies. Additional hallmarks of aging, include genomic instability, chronic inflammation, upregulation of inflammatory cytokines, and deregulated RNA splicing. Prior work demonstrated age-associated splice isoform expression in aged bone marrow (ABM) derived HSPCs (Crews⋯Jamieson, Cell Stem Cell, 2016). Here, we apply a 3D biosensing nanobioreactor (Pham⋯Jamieson, Cell Stem Cell, 2025) to track RNA splicing in aged bone marrow (ABM) derived HSPCs in real-time using confocal microscopy and evaluate age-related splice dysregulation linked to malignant evolution. Methods: Nanobioreactors were constructed from transparent, gas-permeable, 2-port fluoroethyl polymer (FEP) film bags with a porcine gelatin sponge matrix. Mononuclear ABM cells were isolated by Ficoll density centrifugation, and CD34+ cells were purified by magnetic bead selection. CD34+ cells were lentivirally transduced with a dual fluorescent splicing reporter (van der Werf et al., Cell Reports Medicine, 2023), cultured for 48 hours, and co-cultured with autologous CD34- stromal cells in the nanobioreactor at a 1:4 ratio. Results: In ABM-derived CD34+ cells, lentiviral transduction of the dual fluorescent splicing reporter resulted predominantly in red fluorescence protein (RFP) expression, indicating exon inclusion, as assessed by confocal microscopy. Based on preliminary whole-transcriptome sequencing analyses showing increased exon-skipping events, we hypothesize that APOBEC3C drives exon skipping in aged HSPCs. To test this, CD34+ selected cells from ABM aspirates were lentivirally transduced with the splicing reporter, followed by either APOBEC3C overexpression or pCDH vector control. Confocal imaging revealed a fluorescence shift consistent with increased exon skipping in APOBEC3C-transduced cells, confirming the reporter’s ability to capture dynamic splicing changes in response to APOBEC3C activity. Conclusion: We developed a 3D biosensing niche nanobioreactor system that enables live-cell visualization of RNA splicing dynamics in primary human HSPCs. Using this platform, we identified predominant exon inclusion in aged bone marrow-derived HSPCs and demonstrated that APOBEC3C overexpression induces exon skipping, a molecular signature associated with leukemic transformation. Future studies will compare aged and young bone marrow to delineate age-related splicing alterations and define RNA splicing-based biomarkers predictive of hematopoietic aging and malignant evolution. Citation Format: Emma Klacking, Inge van der Werf, Jane Isquith, Jessica Pham, Thomas Whisenant, Anna A. Khachatrian, Ludmil B. Alexandrov, Catriona H. M. Jamieson. Live cell visualization of RNA splicing dynamics upon APOBEC3 lentiviral transduction in aged bone marrow using a 3D biosensing nanobioreactor [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 1387.

  PublisherDOI

• Abstract LB401: The genomic landscape of likely human precancers

  Cancer Research · 2026-04-17

  articleSenior author

  Abstract Precancers represent a diverse collection of lesions that occupy an intermediate state between normal tissues and malignant tumors, yet their biological and clinical definition remains unsettled. To provide a genomic foundation for refining this continuum, we analyzed 1,495 precancers across 17 tissue types using uniformly processed whole-genome and whole-exome sequencing data. This pan-tissue resource reveals striking heterogeneity in the mutational and structural features of precancers, ranging from genomically quiet lesions with minimal alterations to highly aberrant genomes resembling those of invasive cancers. Mutational signature analysis identifies 27 single-base substitution signatures, all previously observed in cancer, indicating that the mutational processes active in malignant cancer are often active at the earliest stages of tumorigenesis. Driver-gene analyses revealed 100 unique genes under positive selection, with TP53 and CDKN2A emerging as the most frequently inactivated, often via biallelic “double-hit” events preceding invasion. Comparative analyses with corresponding cancers showed a progressive enrichment of copy-number alterations, driver mutations, and mutational signatures associated with defective DNA repair and exogenous carcinogens. Random-forest models highlight that cancers converge on a state of extensive genomic disruption, irrespective of the particular mutational processes involved, indicating that canonical tissue-specific driver phenotypes emerge from a backdrop of generic accumulation of genomic alterations that distinguish malignant transformation across tissues. Collectively, this unified atlas of human precancers clarifies the genomic transitions from benign to malignant states, and suggests that many canonical “cancer drivers” may act primarily as general fitness-enhancing alterations detectable well before invasion. These findings provide a genomic framework to support emerging efforts toward molecularly informed early detection and precision prevention strategies in oncology. Citation Format: Christopher D. Steele, Yudou He, Scott M. Lippman, Ludmil B. Alexandrov. The genomic landscape of likely human precancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr LB401.

  PublisherDOI

• Abstract 7631: Predicting and preventing cancer stemness in low Earth orbit

  Cancer Research · 2026-04-03

  article

  Abstract Cancer stemness properties, including enhanced survival, malignant regeneration, telomere deregulation, genomic and epitranscriptomic instability, fuel metastases, and have been linked to stress, retrotransposon and inflammatory cytokine activation, which can occur in low earth orbit (LEO). In NASA Axiom 1, 2 and 3 missions to the ISS, confocal imaging, WGS, RNA-seq and scRNA-seq of lentiviral FUCCI2BL cell cycle and ADAR1-GFP reporter transduced erythroleukemia (TF-1a), colorectal (Caco-2) and metastatic breast cancer (MBC; MDA-MB-231 and patient samples) revealed proliferation, significant genomic instability, HERV and LINE-1 retrotransposon deregulation, and APOBEC3C and ADAR1 activation. Moreover, in Axiom 2 and 3 missions with ADAR1-reporter expressing MBC organoids and in humanized MBC mouse models, an ADAR1p150 splicing modulator, rebecsinib (IND 153126), prevented tumor propagation. Thus, cancer studies in LEO may accelerate the development of innovative cancer therapeutics and countermeasures for long-term spaceflight. Citation Format: Jessica Pham, Wenxue Ma, Claire Engstrom, Patrick Chang, Shuvro P. Nandi, Inge van der Werf, Emma Klacking, Teresa Sposito, Kendale Wirtjes, Thomas Frias, Antonio Ruiz, Jane Isquith, Luisa Ladel, Christina N. Wu, Jana Stoudemire, Pinar Mesci, Kay T. Yeung, Rebecca A. Shatsky, Anna A. Khachatrian, James J. La Clair, Michael D. Burkart, Peggy Wentworth, Curtis L. Scribner, Sheldon R. Morris, Thomas Whisenant, Karla Mack, Ludmil B. Alexandrov, Catriona H. Jamieson. Predicting and preventing cancer stemness in low Earth orbit [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 7631.

  PublisherDOI

• Abstract 1046: Leveraging transcriptomic profiles and deep learning to detect homologous recombination deficiency in breast cancer with softHRD

  Cancer Research · 2026-04-03

  articleSenior author

  Abstract The homologous recombination (HR) pathway is the canonical repair mechanism by which cells repair DNA double-strand breaks. Defects in this pathway, known as homologous recombination deficiency (HRD), can lead to genomic instability and are observed in approximately 13% of breast cancers. HRD is often driven by somatic or germline mutations in BRCA1/2, which render tumors sensitive to PARP inhibitors and platinum-based therapies through synthetic lethality. However, many HRD-positive cancers lack BRCA1/2 mutations, underscoring the need for more reliable approaches to identify tumors likely to benefit from these treatments. To address this, we developed softHRD, a transcriptomics-based framework for detecting HRD in breast cancer. softHRD was trained on RNA-seq profiles from 857 breast cancer patients in The Cancer Genome Atlas (TCGA), filtered for protein-coding genes. A variational autoencoder was first used to reconstruct these transcriptomic profiles, generating latent representations that capture the underlying structure of gene expression patterns. A sparse autoencoder was then applied to these latent features to derive mechanistically interpretable components and identify an HRD-associated gene set. These genes were subsequently leveraged to train a downstream Elastic Net regression model, yielding a robust 111-gene transcriptional signature indicative of HRD. We validated softHRD in 80 breast cancer patients from the I-SPY 2 clinical trial treated with neoadjuvant chemotherapy and olaparib. The model identified a statistically significant difference in pathologic complete response between HRD-predicted and HR-proficient tumors (p = 0.00676). Unlike whole-genome sequencing, which provides a static view of mutational alterations, transcriptomic profiling captures the dynamic state of gene expression, revealing biological changes that genomic methods may overlook. softHRD demonstrated robust performance across all PAM50 breast cancer subtypes, highlighting its generalizability. With the growing integration of transcriptomics into clinical research and diagnostics, softHRD represents a scalable and adaptable framework for accurate, efficient HRD characterization, with potential applications across multiple cancer types. Citation Format: Yashwin Madakamutil, Leo Joseph, Daniyal Rahman, Ammal Abbasi, Ludmil Alexandrov. Leveraging transcriptomic profiles and deep learning to detect homologous recombination deficiency in breast cancer with softHRD [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 1046.

  PublisherDOI

• CarD-T: LLM Automated Literature Review for the Nomination & Analysis of Potential Human Carcinogens

  Zenodo (CERN European Organization for Nuclear Research) · 2026-04-13

  articleOpen access

  Carcinogenic Determination via Transformers (or CarD-T) is an automated pipeline that combines transformer-based machine learning with probabilistic analysis to identify potential carcinogens from biomedical literature. The framework processes accumulating scientific publications (left), applies a trained Named Entity Recognition (NER) model to extract potential carcinogenic entities (center), and Probabilistic Carcinogen Denomination (or PCarD) to analyze temporal trends in evidence shifts (right). This approach enables classification of candidates through Bayesian temporal analysis, overcoming limitations of traditional manual literature review methods.

  PublisherDOI

• Abstract 7205: A universal duplex sequencing approach for accurate detection of somatic mutations

  Cancer Research · 2026-04-03

  articleSenior author

  Abstract Somatic mutations arise from endogenous and exogenous mutagenic processes, accumulating over time and contributing to aging and disease. Detecting these rare mutations in non-clonal tissues remains a significant challenge due to the high error rates, limited genome coverage, and substantial DNA input requirements of existing sequencing approaches. Here, we introduce UDSeq, a high-accuracy, cost-effective, single-molecule duplex sequencing protocol designed to overcome these limitations. We place UDSeq in the context of existing duplex sequencing approaches, demonstrating that it achieves an exceptionally low error rate of ∼2.5 × 10-9 per base pair, supports whole-genome and targeted capture sequencing from as little as 100 picograms of DNA, and delivers up to four times more usable duplex molecules than current state-of-the-art methods from the same input. We demonstrated the broad applicability of UDSeq through a series of in vitro and in vivo mutagenesis experiments, accurately capturing known mutational signatures induced by environmental carcinogens in human cell lines, rodents, and non-model organisms. We further applied UDSeq to normal tissues from a 70-year-old individual, revealing organ-specific mutational burdens and the activity of distinct mutational processes. With its high accuracy, low input requirements, and wide applicability, UDSeq provides a powerful and scalable tool for studying mutational processes across diverse biological contexts. Its versatility supports applications in cancer research, aging, and environmental exposure, expanding our capacity to characterize somatic mutations in both healthy and diseased tissues. Citation Format: Shuvro Prokash Nandi, Yuhe Cheng, Shams Al-azzam, Safa Saeed, Isabella R Stuewe, Zichen Jiang, Luka Culibrk, Maria Zhivagui, Xiaoxu Yang, Rachel M. Wise, Foster C. Jacobs, Bérénice Chavanel, Michael Korenjak, Mia PETLJAK, Silvia Balbo, Laurie G. Hudson, Ke Jian Liu, Jiri Zavadil, Joseph G. Gleeson, Ludmil B. Alexandrov. A universal duplex sequencing approach for accurate detection of somatic mutations [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 7205.

  PublisherDOI

Recent grants

• Comprehensive identification of germline-somatic interactions

  NIH · $1.8M · 2022–2027

• Detecting Mutational Signatures of Environmental Mutagens in Heathy Individuals for Personalized Cancer Prevention

  NIH · $3.4M · 2021–2025

Frequent coauthors

• David C. Wedge

  Cancer Research UK Manchester Institute

  176 shared

• Erik N. Bergstrom

  University of California, San Diego

  141 shared

• Michael R. Stratton

  Wellcome Sanger Institute

  133 shared

• Peter Van Loo

  The University of Texas MD Anderson Cancer Center

  121 shared

• Nischalan Pillay

  University College London

  112 shared

• Peter J. Campbell

  University of Cambridge

  109 shared

• Serena Nik‐Zainal

  108 shared

• Keiran Raine

  Wellcome Sanger Institute

  104 shared

Labs

• The Alexandrov LabPI

Awards & honors

• NIEHS 2020 Outstanding New Environmental Scientist Award
• David and Lucile Packard Foundation 2019 Packard Fellowship…
• V Foundation for Cancer Research 2019 Abeloff V Scholar
• The International Academy for Medical and Biological Enginee…
• Genetics Society 2018 The Balfour Prize Lecture