About

Sriram Sankararaman is an Assistant Professor of Human Genetics at UCLA Samueli School of Engineering, specializing in computational biology, statistical genomics, statistical machine learning, probabilistic graphical models, and Bayesian statistics. He earned his PhD from UC Berkeley in 2010. His research focuses on analyzing DNA to understand human evolution, including the study of ancient human populations such as Neanderthals and other extinct species. His work involves detecting 'ghost' DNA from mysterious human ancestors in modern populations, particularly in West Africans, and exploring how ancient interbreeding events have influenced present-day human genetics. Sankararaman's contributions include uncovering evidence of ancient human admixture and the presence of 'ghost' populations in the genomes of contemporary humans, advancing our understanding of human evolutionary history.

Research topics

• Genetics
• Demography
• Evolutionary biology
• Biology
• Computer Science
• Mathematics
• Statistics
• Geography
• Zoology
• Econometrics

Selected publications

• keju: powerful and accurate inference in Massively Parallel Reporter Assays

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-02-26

  articleOpen access

  Massively Parallel Reporter Assays (MPRAs) interrogate the regulatory function of thousands of designed genetic elements in parallel through linked DNA and RNA readouts using an engineered construct and attached minimal reporter. Given the complexity of MPRA experimental designs, several different sources of uncertainty complicate inference. We show that previous methods do not account for substantial differences in uncertainty levels between the DNA and RNA counts and between batches. Accordingly, we present keju , a hierarchical statistical model that estimates candidate transcription rate, differential activity between conditions, and effects from promoter composition for MPRA data. To maximize statistical power and improve false positive rate control, keju conditions on the DNA counts to model batch-specific and modality-specific uncertainty in the RNA counts. keju shows vastly improved sensitivity (59%) in simulations compared to previous methods (31% for MPRAnalyze and 9% for BCalm), and also has lower, more robust false positive rates, calling only 6.8% of unlabeled negative controls significant in real data (compared to 34% for MPRAnalyze and 12% for BCalm).

  PublisherDOI

• Leveraging ancestral recombination graphs for scalable mixed-model analysis of complex traits

  Cell Genomics · 2025-12-10 · 1 citations

  articleOpen access

  Recent algorithmic advances have enabled the inference of genome-wide ancestral recombination graphs (ARGs) from large genomic cohorts, providing detailed models of genealogical relatedness along the genome. These inferred ARGs can complement genotype imputation by capturing the effects of unobserved variants, but their use in large-scale linear mixed-model analyses has been computationally prohibitive. Here, we develop methods that leverage the ARG to perform genotype-matrix multiplications in sublinear time and implement scalable randomized algorithms for mixed-model analyses. We introduce ARG-RHE, a randomized Haseman-Elston approach for estimating narrow-sense heritability and performing region-based association testing using ARGs, enabling parallel analysis of multiple quantitative traits. Through extensive simulations, we demonstrate the computational efficiency and statistical power of this approach. Applied to 21,159 genes and 52 blood traits in 337,464 UK Biobank participants, ARG-RHE identifies 8% more gene-trait associations than imputation alone, demonstrating that genome-wide genealogies may be leveraged to complement genotype imputation in complex trait analyses.

  PublisherOA PDFDOI

• Investigating the sources of variable impact of pathogenic variants in monogenic metabolic conditions

  Nature Communications · 2025-06-05 · 3 citations

  articleOpen access

  Over three percent of people carry a dominant pathogenic variant, yet only a fraction of carriers develop disease. Disease phenotypes from carriers of variants in the same gene range from mild to severe. Here, we investigate underlying mechanisms for this heterogeneity: variable variant effect sizes, carrier polygenic backgrounds, and modulation of carrier effect by genetic background (marginal epistasis). We leveraged exomes and clinical phenotypes from the UK Biobank and the Mt. Sinai BioMe Biobank to identify carriers of pathogenic variants affecting cardiometabolic traits. We employed recently developed methods to study these cohorts, observing strong statistical support and clinical translational potential for all three mechanisms of variable carrier penetrance and disease severity. For example, scores from our recent model of variant pathogenicity were tightly correlated with phenotype amongst clinical variant carriers, they predicted effects of variants of unknown significance, and they distinguished gain- from loss-of-function variants. We also found that polygenic scores modify phenotypes amongst pathogenic carriers and that genetic background additionally alters the effects of pathogenic variants through interactions.

  PublisherOA PDFDOI

• Neanderthal introgressed ancestry reveals human genomic regions enriched with recessive deleterious mutations

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-05-07 · 1 citations

  preprintOpen access

  Negative natural selection on deleterious mutations plays a key role in shaping human genetic variation. Understanding the dominance of deleterious mutations is critical as it can fundamentally impact the rate and efficiency of natural selection, the magnitude of inbreeding depression, and the prevalence and evolution of genetic diseases. Despite its inarguable importance, the dominance effects of mutations remain poorly understood in humans, primarily because existing statistical methods cannot distinguish them from the overall selective effects of mutations. In this work, we take a fundamentally different approach to infer dominance by leveraging the distribution of Neanderthal ancestry across the human genome. We show through simulations that recessive deleterious mutations lead to an increase in archaic introgressed ancestry in the absence of positive selection, contrary to what is expected when deleterious mutations are additive. Leveraging this unique pattern, we develop a machine learning classifier to infer dominance in genomic windows at a megabase resolution, trained on simulations of a human demographic model with Neanderthal introgression using fully recessive or additive mutations. Our method demonstrates robust accuracy at detecting genomic windows containing recessive deleterious mutations, with particularly high power in exon-dense regions. When applied to the non-African populations from the 1000 Genomes Project, we find that approximately 3-9% of the human genome is enriched for recessive mutations with most recessive regions shared across human populations. Furthermore, our method reveals that recessive deleterious mutations are not evenly distributed across the genome: regions enriched for recessive mutations are significantly depleted of haploinsufficient genes and runs of homozygosity, and are enriched with non-additive variants associated with complex traits. Overall, our Neanderthal ancestry-based approach reveals the presence of recessive deleterious mutations in the human genome and suggests that these mutations are found in regions containing genes associated with metabolism and immune-related traits.

  PublisherOA PDFDOI

• Leveraging protein language models to identify complex trait associations with previously inaccessible classes of functional rare variants

  Cell Genomics · 2025-11-19

  articleOpen access

  Protein language models (PLMs) improve variant effect predictions, but their role in gene discovery for complex traits remains unclear. We introduce an allelic series-based regression test that uses PLM-derived variant effect predictions as proxies for effect sizes, identifying ∼46% more associations than standard burden tests. Extending this to isoform-level analysis, we find 26 gene-trait pairs with stronger associations in non-canonical versus canonical transcripts, highlighting isoform-specific effects. Finally, we identify evolutionary plausible variants (EPVs), missense variants assigned higher likelihoods than the wild-type alleles by PLMs, representing 0.45% of missense variants. EPVs show higher allele frequencies than synonymous variants, consistent with differential selection pressures, and are linked to nine traits, including protective associations with low-density lipoprotein (LDL) and bone mineral density. Together, our results demonstrate how PLMs can enhance rare-variant interpretation and gene-trait association discovery in exome data.

  PublisherDOI

• Estimation and mapping of the missing heritability of human phenotypes

  Nature · 2025-11-12 · 19 citations

  articleOpen access
  PublisherOA PDFDOI

• Single-cell DNA methylome and 3D genome atlas of human subcutaneous adipose tissue

  Nature Genetics · 2025-08-20 · 6 citations

  articleOpen access

  The cell-type-level epigenomic landscape of human subcutaneous adipose tissue (SAT) is not well characterized. Here, we elucidate the epigenomic landscape across SAT cell types using snm3C-seq. We find that SAT CG methylation (mCG) displays pronounced hypermethylation in myeloid cells and hypomethylation in adipocytes and adipose stem and progenitor cells, driving nearly half of the 705,063 differentially methylated regions (DMRs). Moreover, TET1 and DNMT3A are identified as plausible regulators of the cell-type-level mCG profiles. Both global mCG profiles and chromosomal compartmentalization reflect SAT cell-type lineage. Notably, adipocytes display more short-range chromosomal interactions, forming complex local 3D genomic structures that regulate transcriptional functions, including adipogenesis. Furthermore, adipocyte DMRs and A compartments are enriched for abdominal obesity genome-wide association study (GWAS) variants and polygenic risk, while myeloid A compartments are enriched for inflammation. Together, we characterize the SAT single-cell-level epigenomic landscape and link GWAS variants and partitioned polygenic risk of abdominal obesity and inflammation to the SAT epigenome.

  PublisherOA PDFDOI

• Epigenetic patient stratification reveals a sub-endotype of type 2 asthma with altered B-cell response

  medRxiv · 2025-09-02

  preprintOpen access

  Despite biomarker-guided treatment strategies, clinical outcomes among patients with type 2 (T2)-high asthma remain heterogeneous, with some patients responding poorly to T2-targeted biologic therapies. We developed a contrastive machine learning method for patient stratification based on whole-blood DNA methylation (DNAm), applying it to pediatric asthma cohorts of Latino (discovery; n=1,016) and African American (replication; n=429) children. The resulting DNAm stratification score revealed a continuum of clinical severity and drug response within the T2-high asthma endotype. Molecular profiling of high-score asthma patients identified eosinophil-specific hypermethylation-validated in an independent Canadian adult cohort using purified eosinophil DNAm-as well as upregulation of canonical T2-associated genes. Transcriptomic analysis of elevated DNAm scores within T2-high patients further uncovered a gene signature linked to B-cell lineage activity, predominantly reflecting plasma cell activity orthogonal to canonical T2 inflammation programs. This defines a previously unrecognized sub-endotype, which we term T2-high asthma with Altered B Cell response (T2ABC). In a randomized controlled trial of the anti-IgE biologic omalizumab in primarily White adult T2-high asthmatic patients (n=300), the T2ABC gene expression signature was prognostic of poor outcomes, including a 24% mean increase in disease exacerbation rates compared to the trial baseline (P=0.004), which could not be explained by treatment or placebo assignment. Patients treated with omalizumab showed better outcomes than patients in the placebo arm within the T2ABC-low group (P=0.019) but not within the T2ABC-high group (P=0.48), suggesting that IgE blockade does not adequately target the pathogenic mechanisms active in T2ABC-high disease. Single-cell transcriptomic analysis demonstrated that the T2ABC signature reflects heightened activity of non-IgE plasma cells, consistent with the presence of additional antibody isotype responses in a form of severe asthma arising within a T2-high immunologic context. Our findings, replicated and validated across four ancestrally and ethnically diverse pediatric and adult cohorts, support the use of DNAm- and transcriptome-based patient stratification to refine drug development, eligibility, and administration strategies for improving precision in T2 asthma therapy.

  PublisherOA PDFDOI

• CACTI: Leveraging Copy Masking and Contextual Information to Improve Tabular Data Imputation

  PubMed · 2025-06-02

  preprintOpen accessSenior author

  gain of 7.8% over the next best method (13.4%, 6.1%, and 5.3% under missing not at random, at random and completely at random, respectively) - across a diverse range of datasets and missingness conditions. Our results highlight the value of leveraging dataset-specific contextual information and missingness patterns to enhance imputation performance. Code is publicly available at github.com/sriramlab/CACTI.

  PublisherOA PDFDOI

• RECOMB 2025 Special Issue

  Genome Research · 2025-12-01

  preprintOpen access1st authorCorresponding

  This year, the conference received 339 full paper submissions, with 55 ultimately accepted after a rigorous peer-review process involving at least three reviewers per paper.

  PublisherOA PDFDOI

Recent grants

• Statistical Models for Dissecting Human Population Admixture and its Role in Evolution and Disease

  NIH · $1.6M · 2017–2024

• CAREER: Flexible and efficient mixed models to infer the genetic architecture of complex phenotypes

  NSF · $702k · 2020–2026

• Statistical methods to infer structure and impact of ancient admixture

  NIH · $249k · 2014–2018

• III: Medium: Scalable Machine Learning for Genome-Wide Association Analyses

  NSF · $993k · 2017–2022

• Statistical methods to infer structure and impact of ancient admixture

  NIH · $249k · 2014–2018

Frequent coauthors

• Eran Halperin

  University of California, Los Angeles

  130 shared

• Bogdan Paşaniuc

  University of California, Los Angeles

  112 shared

• Yi Ding

  University of California, Los Angeles

  100 shared

• David Reich

  Broad Institute

  98 shared

• Esteban G. Burchard

  University of California, San Francisco

  80 shared

• Noah Zaitlen

  University of California, Los Angeles

  78 shared

• Christopher R. Gignoux

  University of Colorado Anschutz Medical Campus

  63 shared

• Celeste Eng

  University of California, San Francisco

  62 shared

Awards & honors

• Sloan Fellow
• Microsoft Investigator Fellow
• NSF CAREER Award (2018)