About

Amal Ahmed is a professor and the Associate Dean of Graduate Programs at Khoury College of Computer Sciences. Her research focuses on programming languages, with particular attention to semantics, type systems, secure compilation, gradual typing, and software contracts. She is committed to advancing the understanding and development of programming language theory and its applications, contributing to the college's leadership in computer science education and research.

Research topics

• Genetics
• Biology
• Medicine
• Evolutionary biology
• Computer Science
• Computational biology
• Internal medicine
• Demography
• Environmental health
• Artificial Intelligence
• Bioinformatics
• Mathematics
• Theoretical computer science
• Data science
• Computer network
• Geography
• Endocrinology
• Geometry
• Computer vision
• Pure mathematics
• Gerontology
• Programming language
• Surgery

Selected publications

• Severe obesity as an oligogenic condition: evidence from 1714 adults seeking treatment in the UK National Health Service

  Human Genomics · 2025-12-19

  articleOpen access

  Previous evidence has established genetics as an important contributing factor to severe (class III) obesity, which is a chronic, relapsing condition, with a high burden of comorbidity and mortality. We therefore designed a custom genotyping array to screen a cohort of UK patients seeking treatment for severe obesity in a cost-effective way. A total of 1,714 participants were genotyped using a custom AXIOM array, focusing on rare (minor allele frequency < 0.01) variants, with CADD-PHRED ≥ 15 in 78 genes known/suspected to cause Mendelian forms of obesity. Concordance analyses of 22 duplicate samples and 66 samples with whole exome sequence data revealed good genotyping reliability. We identified the proportion of study participants who carried, or were homozygous for, rare, predicted-deleterious variants in genes with dominant and recessive modes of inheritance (MOI), respectively. 27% of patients carried relevant mutations consistent with the expected MOI, which was very similar to the rate observed in the UKB 50 K whole exome sequencing dataset. However, the clinical obesity cohort was more likely to carry two or more such variants, in separate genes, than UK Biobank participants (17.1% vs. 13%, p = 0 0.018), which strongly indicates the possibility of oligogenic inheritance. In conclusion, our results provide evidence: that (i) custom genotyping arrays, used with improved algorithms can allow reliable, cost-effective screening for rare genetic variants; (ii) rare mutations in “obesity genes” may be at high prevalence among bariatric patients, as well as in the general population; and (iii) that severe obesity may have an oligogenic pattern of inheritance in some cases.

  PublisherDOI

• Phenome-wide associations of polygenic scores for schizophrenia and major depression in 100,000 Chinese adults

  medRxiv · 2025-03-04

  preprintOpen access

  Abstract China faces significant mental health challenges, with unique associations between mental disorders and other traits observed in its population. Based on summary statistics in East Asian (EAS) and European (EUR) ancestries, we tested associations of polygenic scores (PGS) for schizophrenia and major depression with 254 phenotypes in 100,640 Chinese adults. PGS predicted schizophrenia (R 2 =1.96%-3.49%) and major depression (R 2 =0.19%-0.77%), and were associated with various socio-demographic, lifestyle, and physical factors. Interestingly, EAS-schizophrenia-PGS was inversely associated with smoking initiation, and EAS-depression-PGS was inversely associated with BMI. Opposing genetic correlations between ancestries were observed for smoking-schizophrenia (inverse in EAS; positive in EUR) and BMI-depression (inverse in EAS; positive in EUR). Mendelian Randomisation supported the causality of these relationships in EUR, but multivariable analyses suggested the influence of pleiotropic effects on other related traits. Our study suggests the context specificity of relationships between mental disorders and other traits, highlighting a potential role of sociocultural factors.

  PublisherOA PDFDOI

• Radar2Shape: 3D Shape Reconstruction from High-Frequency Radar using Multiresolution Signed Distance Functions

  ArXiv.org · 2025-11-21

  preprintOpen accessSenior author

  Determining the shape of 3D objects from high-frequency radar signals is analytically complex but critical for commercial and aerospace applications. Previous deep learning methods have been applied to radar modeling; however, they often fail to represent arbitrary shapes or have difficulty with real-world radar signals which are collected over limited viewing angles. Existing methods in optical 3D reconstruction can generate arbitrary shapes from limited camera views, but struggle when they naively treat the radar signal as a camera view. In this work, we present Radar2Shape, a denoising diffusion model that handles a partially observable radar signal for 3D reconstruction by correlating its frequencies with multiresolution shape features. Our method consists of a two-stage approach: first, Radar2Shape learns a regularized latent space with hierarchical resolutions of shape features, and second, it diffuses into this latent space by conditioning on the frequencies of the radar signal in an analogous coarse-to-fine manner. We demonstrate that Radar2Shape can successfully reconstruct arbitrary 3D shapes even from partially-observed radar signals, and we show robust generalization to two different simulation methods and real-world data. Additionally, we release two synthetic benchmark datasets to encourage future research in the high-frequency radar domain so that models like Radar2Shape can safely be adapted into real-world radar systems.

  PublisherOA PDFDOI

• A Practical Guide for Incorporating Symmetry in Diffusion Policy

  ArXiv.org · 2025-05-19

  preprintOpen access

  Recently, equivariant neural networks for policy learning have shown promising improvements in sample efficiency and generalization, however, their wide adoption faces substantial barriers due to implementation complexity. Equivariant architectures typically require specialized mathematical formulations and custom network design, posing significant challenges when integrating with modern policy frameworks like diffusion-based models. In this paper, we explore a number of straightforward and practical approaches to incorporate symmetry benefits into diffusion policies without the overhead of full equivariant designs. Specifically, we investigate (i) invariant representations via relative trajectory actions and eye-in-hand perception, (ii) integrating equivariant vision encoders, and (iii) symmetric feature extraction with pretrained encoders using Frame Averaging. We first prove that combining eye-in-hand perception with relative or delta action parameterization yields inherent SE(3)-invariance, thus improving policy generalization. We then perform a systematic experimental study on those design choices for integrating symmetry in diffusion policies, and conclude that an invariant representation with equivariant feature extraction significantly improves the policy performance. Our method achieves performance on par with or exceeding fully equivariant architectures while greatly simplifying implementation.

  PublisherOA PDFDOI

• Natural selection exerted by historical coronavirus epidemic(s): comparative genetic analysis in China Kadoorie Biobank and UK Biobank

  BMC Genomics · 2025-10-21

  articleOpen accessSenior author

  BACKGROUND: Pathogens have been one of the primary sources of natural selection affecting modern humans. The footprints of historical selection events - "selective sweeps"- can be detected in the genomes of present-day individuals. Previous analyses of 629 samples from the 1000 Genomes Project suggested that an ancient coronavirus epidemic ~ 20,000 years ago drove multiple selective sweeps in the ancestors of present-day East Asians, but not in other worldwide populations. RESULTS: Using a much larger genetic dataset of 76,719 unrelated individuals from each of the China Kadoorie Biobank (CKB) and UK Biobank (UKB) to identify regions of long-range linkage disequilibrium, we further investigated signatures of past selective sweeps and how they reflect previous viral epidemics. Using independently-curated lists of human host proteins which interact physically or functionally with viruses (virus-interacting proteins; VIPs), we found enrichment in CKB for regions of long-range linkage disequilibrium at genes encoding VIPs for coronaviruses, but not DNA viruses. By contrast, we found no clear evidence for any VIP enrichment in UKB. These findings were supported by additional analyses using saltiLASSI, a selection-scan method robust to false positives caused by demographic events. By contrast, for GWAS signals for SARS-CoV-2 susceptibility (critical illness, hospitalization, and reported infection), there was no difference between UKB and CKB in the number located at or near signals of selection, as expected for a novel virus which has had no opportunity to impact the CKB/UKB study populations. CONCLUSIONS: Together, these results provide evidence of selection events consistent with historical coronavirus epidemic(s) originating in East Asia. These results show how biobank-scale datasets and evolutionary genomics theory can provide insight into the study of past epidemics. The results also highlight how historic infectious disease epidemics can shape the genetic architecture of present-day human populations.

  PublisherOA PDFDOI

• The Ethnic/Racial Variations of Intracerebral Hemorrhage Genetics (ERICH-GENE) Study Protocol

  medRxiv · 2025-06-13 · 1 citations

  preprintOpen access

  Background: Spontaneous, non-traumatic intracranial hemorrhage (ICH) is highly heritable disease. However, the identification of the genetic risk factors driving this high genetic predisposition has been limited by small sample sizes and underrepresentation of non-European populations. The ERICH-GENE study will gather and harmonize clinical, neuroimaging and genomic data on the largest and more diverse collection of ICH cases assembled to date. Methods: ERICH-GENE is an NIH-funded, multi-center, international, genetic and neuroimaging study that aims to achieve the necessary sample size and diversity required to accurately describe the genetic architecture and trans-ethnic variation of ICH. ERICH-GENE will collect and harmonize clinical, neuroimaging and genomic data at least 10,000 multi-ethnic ICH cases. These data will be aggregated with 20,000 existing ICH cases and 600,000 ICH-free controls available through completed studies by the International Stroke Genetics Consortium. To ensure validity, data will undergo extensive harmonization, including expert review of neuroimages to ensure spontaneous etiology and hemorrhage location. We will conduct genome-wide association studies of risk, severity and outcome of ICH, testing for effect modification by race/ethnicity, sex and hemorrhage location. We will also conduct pathway, polygenic risk score and Mendelian randomization analyses. Results: This study will include whole genome sequencing data from 10,850 spontaneous ICH samples, including clinical and radiographic phenotypic data to ensure reliability of true non-traumatic, non-lesional ICH and lobar vs nonlobar location. Of these, 1,497 have already been genotyped using genome-wide arrays, 3,753 have undergone whole genome sequencing, and 5,600 will undergo genome-wide genotyping through ERICH-GENE. There are currently 42 contributing sites exceeding study milestone enrollments. 16,175 radiographic studies from 4,974 patients have been uploaded for harmonization to date, including 26% lobar and 64% nonlobar hemorrhages. Neuroimaging assessment will also include grading for white matter hyperintensities, cerebral atrophy, and presence and severity of IVH. Nearly 6,000 ICH cases will complete genotyping by August 2025. Data/material transfer agreements for summary statistics as well as additional samples are on target to meet the study's objectives. Conclusion: ERICH-GENE is the largest trans-ethnic genetic study of ICH conducted to date. Combining a diverse patient population with expert adjudication of neuroimaging data, ERICH-GENE will identify genetic risk loci that drive the high heritability observed for this disease and make a significant contribution to the understanding of the trans-ethnic variation of its genetic architecture.

  PublisherOA PDFDOI

• Abstract P3024: Phenotypic vs. Genetic Mismatch of Body Mass Index and Its Risk Factors: Evidence from the China Kadoorie Biobank and the UK Biobank

  Circulation · 2025-03-11

  article

  Background: The mismatch between phenotypic and genotypic BMI (BMI-PGM) contributes to the sky-rocketing obesity prevalence in both developed and developing countries. However, little is known about BMI-PGM and its risk factors. Methods: We included 86,205 participants who were genotyped and with valid BMI measurements from the China Kadoorie Biobank (CKB). BMI-PGM was calculated for each participant as the difference between the percentile for his/her adjusted BMI and the percentile for his/her adjusted polygenetic risk score of BMI (BMI-PRS), ranging from -100 to 100. A higher BMI-PGM means one’s measured BMI was much greater than his/her genetically-determined BMI. We then categorized participants into three groups according to BMI-PGM quartiles: discordantly low (bottom quartile), concordant (2nd and 3rd quartiles), and discordantly high (top quartile). Potential risk factors of BMI-PGM, including socio-demographic characteristics and lifestyle factors, were tested by multivariate linear regression. External replication analyses were performed among 102,514 UK Biobank (UKB) participants. Results: In both CKB and UKB, BMI-PGM exhibited a symmetric and normal distribution. Across all four BMI categories (underweight, normal weight, overweight, and obese), the concordant BMI-PGM group accounted for approximately 50% of participants. Of interest, over 50% of obese and underweight adults were categorized into the discordantly high BMI-PGM group and the discordantly low BMI-PGM group, respectively. In both cohorts, participants with higher BMI-PGM were more likely to be younger, current smokers, physically inactive, and meat lovers. However, the associations of BMI-PGM with sex, urban-rural status, education level, household income, alcohol consumption, fruit intake, and sleep duration varied between the two cohorts. Conclusions: A new metric was developed to quantify the mismatch between phenotypic and genetic BMI in both East Asian and European adults. The obesogenic environment affects obesity in different populations with both commonalities and distinct differences.

  PublisherDOI

• Circulating Proteomic Signatures of Pulmonary Function: Multi-ancestry Meta-analysis

  American Journal of Respiratory and Critical Care Medicine · 2025-05-01

  article

  Abstract Pulmonary function assesses the physiologic state of the lungs in health and disease and predicts mortality independently of other risk factors. Large genetic and epigenetic studies have identified many genetic loci associated with pulmonary function. However, proteomic profiles related to pulmonary function have been less examined. Here, we interrogated proteomic profiles in relation to pulmonary function in &amp;gt;20,000 participants (71% European, 14% African, and 15% Asian ancestries) from five cohorts (AGES, ALHS, ARIC, CHS, and CKB). Pulmonary function parameters included forced expiratory volume in one second (FEV1) and forced vital capacity (FVC) measured using spirometry and the FEV1/FVC ratio. Abundance of 4979 human protein analytes (targeting 4860 proteins) was assessed using the Slow Off-rate Modified Aptamer (SOMA)Scan™ platform. Study-level association results using robust linear regression with adjustment for potential confounding factors were meta-analyzed using fixed-effect inverse-variance weighting. In this multi-ancestry population, 1055 proteins were significantly differentially abundant (Bonferroni corrected P&amp;lt;1.004x10-5) in relation to one or more pulmonary function parameters. Enriched pathways implicated inflammatory response and organismal injury. Networks constructed using the significant proteins had more protein-protein interactions than expected. Mendelian randomization analysis results provided evidence of potential causality. Many of the identified associations were validated in independent studies (UKB, STRADL, and QBB) using various proteomic assays. Notably, proteins we found to be related to pulmonary function included many also associated with COPD, either in our data or previous studies. In this extensive investigation of circulating proteins and pulmonary function, we identified proteins and genes not implicated in previous genetic and epigenetic studies. The overlap between identified proteins and known drug targets, approved or under investigation for various respiratory diseases and other health conditions, suggests potential opportunities for drug repurposing. Our findings could enhance the understanding of respiratory health and the development of novel biomarkers and therapeutic strategies for respiratory conditions.

  PublisherDOI

• Expanded mosaic chromosomal alterations, frailty, and risks of all-cause and cause-specific mortality among Chinese and the UK adults: evidence from two prospective cohorts

  BMC Medicine · 2025-11-11

  articleOpen access

  BACKGROUND: Mosaic chromosomal alterations (mCAs) served as a novel indicator of genomic aging. We aimed to investigate the association of expanded mCAs (cell fraction ≥ 10%) with all-cause and cause-specific mortality, and to examine the joint effect of expanded mCAs and frailty index (FI), an indicator of phenotypic aging, on mortality in two large prospective cohorts. METHODS: A total of 100,237 participants in the China Kadoorie Biobank (CKB) and 456,283 participants in the UK Biobank (UKB) were included, followed till Dec 31, 2023, and Nov 30, 2022, respectively. MoChA pipeline was used to detect expanded mCAs events and the subtypes. FIs were calculated using previously validated equations, with 28 items included in the CKB and 49 items in the UKB, and categorized participants into three groups: robust, prefrail, and frail. Adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated to examine the associations of the expanded mCAs and joint categories of frailty-mCAs with all-cause and cause-specific mortality by using Cox proportional hazards models. The combined effect values of two cohorts were estimated using random-effects models by meta-analysis. RESULTS: The prevalence of expanded mCAs in the CKB and UKB was 2.2% and 3.4%, respectively. After a median follow-up of 17.2 years in the CKB and 13.7 years in the UKB, expanded mCAs carriers had a higher risk of all-cause (HRs [95% CIs]: 1.20 [1.16, 1.24]) and risks of cause-specific mortality (HRs [95% CIs]: 1.27 [1.21, 1.34], 1.13 [1.02, 1.25], and 1.24 [1.12, 1.37] for death from cancers, circulatory diseases, and respiratory diseases, respectively). Such associations largely did not overlap with FI, especially for all-cause and cancer mortality. Joint analyses revealed that individuals with lower frailty level but with expanded mCAs had a comparable and even higher risk of cancer mortality compared to those with higher frailty level but without mCAs. Similar pattern was also found in terms of adjusted 10-year cancer mortality rates. CONCLUSIONS: Our findings suggested that expanded mCAs were significantly associated with all-cause and cause-specific deaths and could serve as a complement to the FI in providing a more comprehensive perspective on mortality risk, especially for cancer mortality.

  PublisherOA PDFDOI

• Causal Relevance of Lp(a) for Coronary Heart Disease and Stroke Types in East Asian and European Ancestry Populations: A Mendelian Randomization Study

  Circulation · 2025-04-29 · 14 citations

  articleOpen access

  BACKGROUND: Elevated plasma levels of Lp(a) [lipoprotein(a)] are a causal risk factor for coronary heart disease and stroke in European individuals, but the causal relevance of Lp(a) for different stroke types and in East Asian individuals with different Lp(a) genetic architecture is uncertain. METHODS: We measured plasma levels of Lp(a) in a nested case-control study of 18 174 adults (mean [SD] age, 57 [10] years; 49% female) in the China Kadoorie Biobank (CKB) and performed a genome-wide association analysis to identify genetic variants affecting Lp(a) levels, with replication in ancestry-specific subsets in UK Biobank. We further performed 2-sample Mendelian randomization analyses, associating ancestry-specific Lp(a)-associated instrumental variants derived from CKB or from published data in European individuals with risk of myocardial infarction (n=17 091), ischemic stroke (IS [n=29 233]) and its subtypes, or intracerebral hemorrhage (n=5845) in East Asian and European individuals using available data from CKB and genome-wide association analysis consortia. RESULTS: In CKB observational analyses, plasma levels of Lp(a) were log-linearly and positively associated with higher risks of myocardial infarction and IS, but not with intracerebral hemorrhage. In genome-wide association analysis, we identified 29 single nucleotide polymorphisms independently associated with Lp(a) that together explained 33% of variance in Lp(a) in Chinese individuals. In UK Biobank, the lead Chinese variants identified in CKB were replicated in 1260 Chinese individuals, but explained only 10% of variance in Lp(a) in European individuals. In Mendelian randomization analyses, however, there were highly concordant effects of Lp(a) across both ancestries for all cardiovascular disease outcomes examined. In combined analyses of both ancestries, the proportional reductions in risk per 100 nmol/L lower genetically predicted Lp(a) levels for myocardial infarction were 3-fold greater than for total IS (rate ratio, 0.78 [95% CI, 0.76-0.81] versus 0.94 [0.92-0.96]), but were similar to those for large-artery IS (0.80 [0.73-0.87]; n=8134). There were weaker associations with cardioembolic IS (0.92 [95% CI, 0.86-0.98]; n=11 730), and no association with small-vessel IS (0.99 [0.91-1.07]; n=12 343) or with intracerebral hemorrhage (1.08 [0.96-1.21]; n=5845). CONCLUSIONS: The effects of Lp(a) on risk of myocardial infarction and large-artery IS were comparable in East Asian and European individuals, suggesting that people with either ancestry could expect comparable proportional benefits for equivalent reductions in Lp(a), but there was little effect on other stroke types.

  PublisherDOI

Recent grants

• Collaborative Research: SCALE MoDL: Representation Theoretic Foundations of Deep Learning

  NSF · $661k · 2022–2025

Frequent coauthors

• Zhengming Chen

  University of Oxford

  558 shared

• Iona Y. Millwood

  University of Oxford

  527 shared

• Huaidong Du

  Guangzhou Marine Geological Survey

  443 shared

• Jun Lv

  Guizhou University

  435 shared

• Yiping Chen

  Medical Research Council

  426 shared

• Liming Li

  Peking University

  407 shared

• Ling Yang

  University of Oxford

  394 shared

• Canqing Yu

  Peking University

  369 shared

Labs

• Robin Walters LabPI