Honors & Awards


  • NSF Graduate Research Fellow, NSF (2019 - present)
  • Knight-Hennessy Scholar, Stanford University (2018 - present)

Stanford Advisors


All Publications


  • An 8-gene machine learning model improves clinical prediction of severe dengue progression. Genome medicine Liu, Y. E., Saul, S., Rao, A. M., Robinson, M. L., Agudelo Rojas, O. L., Sanz, A. M., Verghese, M., Solis, D., Sibai, M., Huang, C. H., Sahoo, M. K., Gelvez, R. M., Bueno, N., Estupinan Cardenas, M. I., Villar Centeno, L. A., Rojas Garrido, E. M., Rosso, F., Donato, M., Pinsky, B. A., Einav, S., Khatri, P. 2022; 14 (1): 33

    Abstract

    BACKGROUND: Each year 3-6 million people develop life-threatening severe dengue (SD). Clinical warning signs for SD manifest late in the disease course and are nonspecific, leading to missed cases and excess hospital burden. Better SD prognostics are urgently needed.METHODS: We integrated 11 public datasets profiling the blood transcriptome of 365 dengue patients of all ages and from seven countries, encompassing biological, clinical, and technical heterogeneity. We performed an iterative multi-cohort analysis to identify differentially expressed genes (DEGs) between non-severe patients and SD progressors. Using only these DEGs, we trained an XGBoost machine learning model on public data to predict progression to SD. All model parameters were "locked" prior to validation in an independent, prospectively enrolled cohort of 377 dengue patients in Colombia. We measured expression of the DEGs in whole blood samples collected upon presentation, prior to SD progression. We then compared the accuracy of the locked XGBoost model and clinical warning signs in predicting SD.RESULTS: We identified eight SD-associated DEGs in the public datasets and built an 8-gene XGBoost model that accurately predicted SD progression in the independent validation cohort with 86.4% (95% CI 68.2-100) sensitivity and 79.7% (95% CI 75.5-83.9) specificity. Given the 5.8% proportion of SD cases in this cohort, the 8-gene model had a positive and negative predictive value (PPV and NPV) of 20.9% (95% CI 16.7-25.6) and 99.0% (95% CI 97.7-100.0), respectively. Compared to clinical warning signs at presentation, which had 77.3% (95% CI 58.3-94.1) sensitivity and 39.7% (95% CI 34.7-44.9) specificity, the 8-gene model led to an 80% reduction in the number needed to predict (NNP) from 25.4 to 5.0. Importantly, the 8-gene model accurately predicted subsequent SD in the first three days post-fever onset and up to three days prior to SD progression.CONCLUSIONS: The 8-gene XGBoost model, trained on heterogeneous public datasets, accurately predicted progression to SD in a large, independent, prospective cohort, including during the early febrile stage when SD prediction remains clinically difficult. The model has potential to be translated to a point-of-care prognostic assay to reduce dengue morbidity and mortality without overwhelming limited healthcare resources.

    View details for DOI 10.1186/s13073-022-01034-w

    View details for PubMedID 35346346

  • Factors associated with COVID-19 vaccine acceptance and hesitancy among residents of Northern California jails. Preventive medicine reports Liu, Y. E., Oto, J., Will, J., LeBoa, C., Doyle, A., Rens, N., Aggarwal, S., Kalish, I., Rodriguez, M., Sherif, B., Trinidad, C., Del Rosario, M., Allen, S., Spencer, R., Morales, C., Chyorny, A., Andrews, J. R. 2022; 27: 101771

    Abstract

    Carceral facilities are high-risk settings for COVID-19 transmission. Factors associated with COVID-19 vaccine acceptance and hesitancy among incarcerated individuals are poorly understood, especially among jail residents. Here, we conducted a retrospective review of electronic health record (EHR) data on COVID-19 vaccine uptake in custody and additionally administered a survey to assess reasons for vaccine hesitancy, sources of COVID-19 information, and medical mistrust among residents of four Northern California jails. We performed multivariate logistic regression to determine associations with vaccine acceptance. Of 2,564 jail residents offered a COVID-19 vaccine between March 19, 2021 and June 30, 2021, 1,441 (56.2%) accepted at least one dose. Among vaccinated residents, 497 (34.5%) had initially refused. Vaccine uptake was higher among older individuals, women, those with recent flu vaccination, and those living in shared housing. Among 509 survey respondents, leading reasons for vaccine hesitancy were concerns around side effects and suboptimal efficacy, with cost and the need for an annual booster being other hypothetical deterrents to vaccination. Vaccine hesitancy was also associated with mistrust of medical personnel in and out of jail, although this association varied by race/ethnicity. Television and friends/family were the most common and most trusted sources of COVID-19 information, respectively. Overall, vaccine acceptance was much lower among jail residents than the local and national general population. Interventions to increase vaccination rates in this setting should utilize accessible and trusted sources of information to address concerns about side effects and efficacy, while working to mitigate medical and institutional mistrust among residents.

    View details for DOI 10.1016/j.pmedr.2022.101771

    View details for PubMedID 35309721

    View details for PubMedCentralID PMC8920969

  • All-cause and cause-specific mortality during and following incarceration in Brazil: A retrospective cohort study. PLoS medicine Liu, Y. E., Lemos, E. F., Gonçalves, C. C., de Oliveira, R. D., Santos, A. d., do Prado Morais, A. O., Croda, M. G., de Lourdes Delgado Alves, M., Croda, J., Walter, K. S., Andrews, J. R. 2021; 18 (9): e1003789

    Abstract

    Mortality during and after incarceration is poorly understood in low- and middle-income countries (LMICs). The need to address this knowledge gap is especially urgent in South America, which has the fastest growing prison population in the world. In Brazil, insufficient data have precluded our understanding of all-cause and cause-specific mortality during and after incarceration.We linked incarceration and mortality databases for the Brazilian state of Mato Grosso do Sul to obtain a retrospective cohort of 114,751 individuals with recent incarceration. Between January 1, 2009 and December 31, 2018, we identified 3,127 deaths of individuals with recent incarceration (705 in detention and 2,422 following release). We analyzed age-standardized, all-cause, and cause-specific mortality rates among individuals detained in different facility types and following release, compared to non-incarcerated residents. We additionally modeled mortality rates over time during and after incarceration for all causes of death, violence, or suicide. Deaths in custody were 2.2 times the number reported by the national prison administration (n = 317). Incarcerated men and boys experienced elevated mortality, compared with the non-incarcerated population, due to increased risk of death from violence, suicide, and communicable diseases, with the highest standardized incidence rate ratio (IRR) in semi-open prisons (2.4; 95% confidence interval [CI]: 2.0 to 2.8), police stations (3.1; 95% CI: 2.5 to 3.9), and youth detention (8.1; 95% CI: 5.9 to 10.8). Incarcerated women experienced increased mortality from suicide (IRR = 6.0, 95% CI: 1.2 to 17.7) and communicable diseases (IRR = 2.5, 95% CI: 1.1 to 5.0). Following release from prison, mortality was markedly elevated for men (IRR = 3.0; 95% CI: 2.8 to 3.1) and women (IRR = 2.4; 95% CI: 2.1 to 2.9). The risk of violent death and suicide was highest immediately post-release and declined over time; however, all-cause mortality remained elevated 8 years post-release. The limitations of this study include inability to establish causality, uncertain reliability of data during incarceration, and underestimation of mortality rates due to imperfect database linkage.Incarcerated individuals in Brazil experienced increased mortality from violence, suicide, and communicable diseases. Mortality was heightened following release for all leading causes of death, with particularly high risk of early violent death and elevated all-cause mortality up to 8 years post-release. These disparities may have been underrecognized in Brazil due to underreporting and insufficient data.

    View details for DOI 10.1371/journal.pmed.1003789

    View details for PubMedID 34534214

  • The role of prisons in disseminating tuberculosis in Brazil: A genomic epidemiology study. Lancet Regional Health. Americas Walter, K. S., Dos Santos, P. C., Gonçalves, T. O., da Silva, B. O., da Silva Santos, A., de Cássia Leite, A., da Silva, A. M., Figueira Moreira, F. M., de Oliveira, R. D., Lemos, E. F., Cunha, E., Liu, Y. E., Ko, A. I., Colijn, C., Cohen, T., Mathema, B., Croda, J., Andrews, J. R. 2022; 9

    Abstract

    Globally, prisons are high-incidence settings for tuberculosis. Yet the role of prisons as reservoirs of M. tuberculosis, propagating epidemics through spillover to surrounding communities, has been difficult to measure directly.To quantify the role of prisons in driving wider community M. tuberculosis transmission, we conducted prospective genomic surveillance in Central West Brazil from 2014 to 2019. We whole genome sequenced 1152 M. tuberculosis isolates collected during active and passive surveillance inside and outside prisons and linked genomes to detailed incarceration histories. We applied multiple phylogenetic and genomic clustering approaches and inferred timed transmission trees.M. tuberculosis sequences from incarcerated and non-incarcerated people were closely related in a maximum likelihood phylogeny. The majority (70.8%; 46/65) of genomic clusters including people with no incarceration history also included individuals with a recent history of incarceration. Among cases in individuals with no incarceration history, 50.6% (162/320) were in clusters that included individuals with recent incarceration history, suggesting that transmission chains often span prisons and communities. We identified a minimum of 18 highly probable spillover events, M. tuberculosis transmission from people with a recent incarceration history to people with no prior history of incarceration, occurring in the state's four largest cities and across sampling years. We additionally found that frequent transfers of people between the state's prisons creates a highly connected prison network that likely disseminates M. tuberculosis across the state.We developed a framework for measuring spillover from high-incidence environments to surrounding communities by integrating genomic and spatial information. Our findings indicate that, in this setting, prisons serve not only as disease reservoirs, but also disseminate M. tuberculosis across highly connected prison networks, both amplifying and propagating M. tuberculosis risk in surrounding communities.Brazil's National Council for Scientific and Technological Development and US National Institutes of Health.

    View details for DOI 10.1016/j.lana.2022.100186

    View details for PubMedID 35647574

    View details for PubMedCentralID PMC9140320

  • A 6-mRNA host response classifier in whole blood predicts outcomes in COVID-19 and other acute viral infections. Scientific reports Buturovic, L., Zheng, H., Tang, B., Lai, K., Kuan, W. S., Gillett, M., Santram, R., Shojaei, M., Almansa, R., Nieto, J. A., Munoz, S., Herrero, C., Antonakos, N., Koufargyris, P., Kontogiorgi, M., Damoraki, G., Liesenfeld, O., Wacker, J., Midic, U., Luethy, R., Rawling, D., Remmel, M., Coyle, S., Liu, Y. E., Rao, A. M., Dermadi, D., Toh, J., Jones, L. M., Donato, M., Khatri, P., Giamarellos-Bourboulis, E. J., Sweeney, T. E. 1800; 12 (1): 889

    Abstract

    Predicting the severity of COVID-19 remains an unmet medical need. Our objective was to develop a blood-based host-gene-expression classifier for the severity of viral infections and validate it in independent data, including COVID-19. We developed a logistic regression-based classifier for the severity of viral infections and validated it in multiple viral infection settings including COVID-19. We used training data (N=705) from 21 retrospective transcriptomic clinical studies of influenza and other viral illnesses looking at a preselected panel of host immune response messenger RNAs. We selected 6 host RNAs and trained logistic regression classifier with a cross-validation area under curve of 0.90 for predicting 30-day mortality in viral illnesses. Next, in 1417 samples across 21 independent retrospective cohorts the locked 6-RNA classifier had an area under curve of 0.94 for discriminating patients with severe vs. non-severe infection. Next, in independent cohorts of prospectively (N=97) and retrospectively (N=100) enrolled patients with confirmed COVID-19, the classifier had an area under curve of 0.89 and 0.87, respectively, for identifying patients with severe respiratory failure or 30-day mortality. Finally, we developed a loop-mediated isothermal gene expression assay for the 6-messenger-RNA panel to facilitate implementation as a rapid assay. With further study, the classifier could assist in the risk assessment of COVID-19 and other acute viral infections patients to determine severity and level of care, thereby improving patient management and reducing healthcare burden.

    View details for DOI 10.1038/s41598-021-04509-9

    View details for PubMedID 35042868

  • Multi-cohort analysis of host immune response identifies conserved protective and detrimental modules associated with severity across viruses. Immunity Zheng, H., Rao, A. M., Dermadi, D., Toh, J., Murphy Jones, L., Donato, M., Liu, Y., Su, Y., Dai, C. L., Kornilov, S. A., Karagiannis, M., Marantos, T., Hasin-Brumshtein, Y., He, Y. D., Giamarellos-Bourboulis, E. J., Heath, J. R., Khatri, P. 2021

    Abstract

    Viral infections induce a conserved host response distinct from bacterial infections. We hypothesized that the conserved response is associated with disease severity and is distinct between patients with different outcomes. To test this, we integrated 4,780 blood transcriptome profiles from patients aged 0 to 90 years infected with one of 16 viruses, including SARS-CoV-2, Ebola, chikungunya, and influenza, across 34 cohorts from 18 countries, and single-cell RNA sequencing profiles of 702,970 immune cells from 289 samples across three cohorts. Severe viral infection was associated with increased hematopoiesis, myelopoiesis, and myeloid-derived suppressor cells. We identified protective and detrimental gene modules that defined distinct trajectories associated with mild versus severe outcomes. The interferon response was decoupled from the protective host response in patients with severe outcomes. These findings were consistent, irrespective of age and virus, and provide insights to accelerate the development of diagnostics and host-directed therapies to improve global pandemic preparedness.

    View details for DOI 10.1016/j.immuni.2021.03.002

    View details for PubMedID 33765435

  • Covid-19 Vaccine Acceptance in California State Prisons. The New England journal of medicine Chin, E. T., Leidner, D. n., Ryckman, T. n., Liu, Y. E., Prince, L. n., Alarid-Escudero, F. n., Andrews, J. R., Salomon, J. A., Goldhaber-Fiebert, J. D., Studdert, D. M. 2021

    View details for DOI 10.1056/NEJMc2105282

    View details for PubMedID 33979505

  • Combinatorial ETS1-dependent control of oncogenic NOTCH1 enhancers in T-cell leukemia. Blood cancer discovery McCarter, A. C., Gatta, G. D., Melnick, A., Kim, E., Sha, C., Wang, Q., Nalamolu, J. K., Liu, Y., Keeley, T. M., Yan, R., Sun, M., Kodgule, R., Kunnath, N., Ambesi-Impiombato, A., Kuick, R., Rao, A., Ryan, R. J., Kee, B. L., Samuelson, L. C., Ostrowski, M. C., Ferrando, A. A., Chiang, M. Y. 2020; 1 (2): 178–97

    Abstract

    Notch activation is highly prevalent among cancers, in particular T-cell acute lymphoblastic leukemia (T-ALL). However, the use of pan-Notch inhibitors to treat cancers has been hampered by adverse effects, particularly intestinal toxicities. To circumvent this barrier in T-ALL, we aimed to inhibit ETS1, a developmentally important T-cell transcription factor previously shown to co-bind Notch response elements. Using complementary genetic approaches in mouse models, we show that ablation of Ets1 leads to strong Notch-mediated suppressive effects on T-cell development and leukemogenesis, but milder intestinal effects than pan-Notch inhibitors. Mechanistically, genome-wide chromatin profiling studies demonstrate that Ets1 inactivation impairs recruitment of multiple Notch-associated factors and Notch-dependent activation of transcriptional elements controlling major Notch-driven oncogenic effector pathways. These results uncover previously unrecognized hierarchical heterogeneity of Notch-controlled genes and points to Ets1-mediated enucleation of Notch-Rbpj transcriptional complexes as a target for developing specific anti-Notch therapies in T-ALL that circumvent the barriers of pan-Notch inhibition.

    View details for DOI 10.1158/2643-3230.bcd-20-0026

    View details for PubMedID 32924017

  • Cl-? and H+ coupling properties and subcellular localizations of wildtype and disease-associated variants of the voltage-gated Cl-?/H+ exchanger ClC-5 JOURNAL OF BIOLOGICAL CHEMISTRY Chang, M., Brown, M. R., Liu, Y., Gainullin, V. G., Harris, P. C., Romero, M. F., Lieske, J. C. 2020; 295 (6): 1464–73

    Abstract

    Dent disease 1 (DD1) is caused by mutations in the CLCN5 gene encoding a voltage-gated electrogenic nCl-/H+ exchanger ClC-5. Using ion-selective microelectrodes and Xenopus oocytes, here we studied Cl-/H+ coupling properties of WT ClC-5 and four DD1-associated variants (S244L, R345W, Q629*, and T657S), along with trafficking and localization of ClC-5. WT ClC-5 had a 2Cl-/H+ exchange ratio at a Vh of +40 mV with a [Cl-]out of 104 mm, but the transport direction did not reverse with a [Cl-]out of 5 mm, indicating that ClC-5-mediated exchange of two Cl- out for one H+ in is not permissible. We hypothesized that ClC-5 and H+-ATPase are functionally coupled during H+-ATPase-mediated endosomal acidification, crucial for ClC-5 activation by depolarizing endosomes. ClC-5 transport that provides three net negative charges appeared self-inhibitory because of ClC-5's voltage-gated properties, but shunt conductance facilitated further H+-ATPase-mediated endosomal acidification. Thus, an on-and-off "burst" of ClC-5 activity was crucial for preventing Cl- exit from endosomes. The subcellular distribution of the ClC-5:S244L variant was comparable with that of WT ClC-5, but the variant had a much slower Cl- and H+ transport and displayed an altered stoichiometry of 1.6:1. The ClC-5:R345W variant exhibited slightly higher Cl-/H+ transport than ClC-5:S244L, but co-localized with early endosomes, suggesting decreased ClC-5:R345W membrane trafficking is perhaps in a fully functional form. The truncated ClC-5:Q629* variant displayed the lowest Cl-/H+ exchange and was retained in the endoplasmic reticulum and cis-Golgi, but not in early endosomes, suggesting the nonsense mutation affects ClC-5 maturation and trafficking.

    View details for DOI 10.1074/jbc.RA119.011366

    View details for Web of Science ID 000514326500005

    View details for PubMedID 31852738

    View details for PubMedCentralID PMC7008381

  • Loss of Extreme Long-Range Enhancers in Human Neural Crest Drives a Craniofacial Disorder. Cell stem cell Long, H. K., Osterwalder, M. n., Welsh, I. C., Hansen, K. n., Davies, J. O., Liu, Y. E., Koska, M. n., Adams, A. T., Aho, R. n., Arora, N. n., Ikeda, K. n., Williams, R. M., Sauka-Spengler, T. n., Porteus, M. H., Mohun, T. n., Dickel, D. E., Swigut, T. n., Hughes, J. R., Higgs, D. R., Visel, A. n., Selleri, L. n., Wysocka, J. n. 2020

    Abstract

    Non-coding mutations at the far end of a large gene desert surrounding the SOX9 gene result in a human craniofacial disorder called Pierre Robin sequence (PRS). Leveraging a human stem cell differentiation model, we identify two clusters of enhancers within the PRS-associated region that regulate SOX9 expression during a restricted window of facial progenitor development at distances up to 1.45 Mb. Enhancers within the 1.45 Mb cluster exhibit highly synergistic activity that is dependent on the Coordinator motif. Using mouse models, we demonstrate that PRS phenotypic specificity arises from the convergence of two mechanisms: confinement of Sox9 dosage perturbation to developing facial structures through context-specific enhancer activity and heightened sensitivity of the lower jaw to Sox9 expression reduction. Overall, we characterize the longest-range human enhancers involved in congenital malformations, directly demonstrate that PRS is an enhanceropathy, and illustrate how small changes in gene expression can lead to morphological variation.

    View details for DOI 10.1016/j.stem.2020.09.001

    View details for PubMedID 32991838

  • Stage-specific roles for Zmiz1 in Notch-dependent steps of early T-cell development. Blood Wang, Q., Yan, R., Pinnell, N., McCarter, A. C., Oh, Y., Liu, Y., Sha, C., Garber, N. F., Chen, Y., Wu, Q., Ku, C. J., Tran, I., Serna Alarcon, A., Kuick, R., Engel, J. D., Maillard, I., Cierpicki, T., Chiang, M. Y. 2018; 132 (12): 1279-1292

    Abstract

    Notch1 signaling must elevate to high levels in order to drive the proliferation of CD4-CD8- double-negative (DN) thymocytes and progression to the CD4+CD8+ double-positive (DP) stage through β-selection. During this critical phase of pre-T-cell development, which is also known as the DN-DP transition, it is unclear whether the Notch1 transcriptional complex strengthens its signal output as a discrete unit or through cofactors. We previously showed that the protein inhibitor of activated STAT-like coactivator Zmiz1 is a context-dependent cofactor of Notch1 in T-cell leukemia. We also showed that withdrawal of Zmiz1 generated an early T-lineage progenitor (ETP) defect. Here, we show that this early defect seems inconsistent with loss-of-Notch1 function. In contrast, at the later pre-T-cell stage, withdrawal of Zmiz1 impaired the DN-DP transition by inhibiting proliferation, like withdrawal of Notch. In pre-T cells, but not ETPs, Zmiz1 cooperatively regulated Notch1 target genes Hes1, Lef1, and Myc. Enforced expression of either activated Notch1 or Myc partially rescued the Zmiz1-deficient DN-DP defect. We identified residues in the tetratricopeptide repeat (TPR) domain of Zmiz1 that bind Notch1. Mutating only a single residue impaired the Zmiz1-Notch1 interaction, Myc induction, the DN-DP transition, and leukemic proliferation. Similar effects were seen using a dominant-negative TPR protein. Our studies identify stage-specific roles of Zmiz1. Zmiz1 is a context-specific cofactor for Notch1 during Notch/Myc-dependent thymocyte proliferation, whether normal or malignant. Finally, we highlight a vulnerability in leukemic cells that originated from a developmentally important Zmiz1-Notch1 interaction that is hijacked during transformation from normal pre-T cells.

    View details for DOI 10.1182/blood-2018-02-835850

    View details for PubMedID 30076146

    View details for PubMedCentralID PMC6148450

  • The PIAS-like Coactivator Zmiz1 Is a Direct and Selective Cofactor of Notch1 in T Cell Development and Leukemia. Immunity Pinnell, N., Yan, R., Cho, H. J., Keeley, T., Murai, M. J., Liu, Y., Alarcon, A. S., Qin, J., Wang, Q., Kuick, R., Elenitoba-Johnson, K. S., Maillard, I., Samuelson, L. C., Cierpicki, T., Chiang, M. Y. 2015; 43 (5): 870-83

    Abstract

    Pan-NOTCH inhibitors are poorly tolerated in clinical trials because NOTCH signals are crucial for intestinal homeostasis. These inhibitors might also promote cancer because NOTCH can act as a tumor suppressor. We previously reported that the PIAS-like coactivator ZMIZ1 is frequently co-expressed with activated NOTCH1 in T cell acute lymphoblastic leukemia (T-ALL). Here, we show that similar to Notch1, Zmiz1 was important for T cell development and controlled the expression of certain Notch target genes, such as Myc. However, unlike Notch, Zmiz1 had no major role in intestinal homeostasis or myeloid suppression. Deletion of Zmiz1 impaired the initiation and maintenance of Notch-induced T-ALL. Zmiz1 directly interacted with Notch1 via a tetratricopeptide repeat domain at a special class of Notch-regulatory sites. In contrast to the Notch cofactor Maml, which is nonselective, Zmiz1 was selective. Thus, targeting the NOTCH1-ZMIZ1 interaction might combat leukemic growth while avoiding the intolerable toxicities of NOTCH inhibitors.

    View details for DOI 10.1016/j.immuni.2015.10.007

    View details for PubMedID 26522984

    View details for PubMedCentralID PMC4654973