Dictionary learning for integrative, multimodal and scalable single-cell analysis

Mapping single-cell sequencing profiles to comprehensive reference datasets provides a powerful alternative to unsupervised analysis. However, most reference datasets are constructed from single-cell RNA-sequencing data and cannot be used to annotate datasets that do not measure gene expression. Here we introduce 'bridge integration', a method to integrate single-cell datasets across modalities using a multiomic dataset as a molecular bridge. Each cell in the multiomic dataset constitutes an element in a 'dictionary', which is used to reconstruct unimodal datasets and transform them into a shared space. Our procedure accurately integrates transcriptomic data with independent single-cell measurements of chromatin accessibility, histone modifications, DNA methylation and protein levels. Moreover, we demonstrate how dictionary learning can be combined with sketching techniques to improve computational scalability and harmonize 8.6 million human immune cell profiles from sequencing and mass cytometry experiments. Our approach, implemented in version 5 of our Seurat toolkit ( http://www.satijalab.org/seurat ), broadens the utility of single-cell reference datasets and facilitates comparisons across diverse molecular modalities.

CPA-Perturb-seq: Multiplexed single-cell characterization of alternative polyadenylation regulators

Most mammalian genes have multiple polyA sites, representing a substantial source of transcript diversity that is governed by the cleavage and polyadenylation (CPA) regulatory machinery. To better understand how these proteins govern polyA site choice we introduce CPA-Perturb-seq, a multiplexed perturbation screen dataset of 42 known CPA regulators with a 3' scRNA-seq readout that enables transcriptome-wide inference of polyA site usage. We develop a statistical framework to specifically identify perturbation-dependent changes in intronic and tandem polyadenylation, and discover modules of co-regulated polyA sites exhibiting distinct functional properties. By training a multi-task deep neural network (APARENT-Perturb) on our dataset, we delineate a cis-regulatory code that predicts responsiveness to perturbation and reveals interactions between distinct regulatory complexes. Finally, we leverage our framework to re-analyze published scRNA-seq datasets, identifying new regulators that affect the relative abundance of alternatively polyadenylated transcripts, and characterizing extensive cellular heterogeneity in 3' UTR length amongst antibody-producing cells. Our work highlights the potential for multiplexed single-cell perturbation screens to further our understanding of post-transcriptional regulation in vitro and in vivo.

Increased tryptophan, but not increased glucose metabolism, predict resistance of pembrolizumab in stage III/IV melanoma

Clinical trials of combined IDO/PD1 blockade in metastatic melanoma (MM) failed to show additional clinical benefit compared to PD1-alone inhibition. We reasoned that a tryptophan-metabolizing pathway other than the kynurenine one is essential. We immunohistochemically stained tissues along the nevus-to-MM progression pathway for tryptophan-metabolizing enzymes (TMEs; TPH1, TPH2, TDO2, IDO1) and the tryptophan transporter, LAT1. We assessed tryptophan and glucose metabolism by performing baseline C11-labeled α-methyl tryptophan (C11-AMT) and fluorodeoxyglucose (FDG) PET imaging of tumor lesions in a prospective clinical trial of pembrolizumab in MM (clinicaltrials.gov, NCT03089606). We found higher protein expression of all TMEs and LAT1 in melanoma cells than tumor-infiltrating lymphocytes (TILs) within MM tumors (n = 68). Melanoma cell-specific TPH1 and LAT1 expressions were significantly anti-correlated with TIL presence in MM. High melanoma cell-specific LAT1 and low IDO1 expression were associated with worse overall survival (OS) in MM. Exploratory optimal cutpoint survival analysis of pretreatment 'high' vs. 'low' C11-AMT SUV_max of the hottest tumor lesion per patient revealed that the 'low' C11-AMT SUV_max was associated with longer progression-free survival in our clinical trial (n = 26). We saw no such trends with pretreatment FDG PET SUV_max. Treatment of melanoma cell lines with telotristat, a TPH1 inhibitor, increased IDO expression and kynurenine production in addition to suppression of serotonin production. High melanoma tryptophan metabolism is a poor predictor of pembrolizumab response and an adverse prognostic factor. Serotoninergic but not kynurenine pathway activation may be significant. Melanoma cells outcompete adjacent TILs, eventually depriving the latter of an essential amino acid.

Transcriptome-Wide Association Study of Blood Cell Traits in African Ancestry and Hispanic/Latino Populations

BACKGROUND

Thousands of genetic variants have been associated with hematological traits, though target genes remain unknown at most loci. Moreover, limited analyses have been conducted in African ancestry and Hispanic/Latino populations; hematological trait associated variants more common in these populations have likely been missed.

METHODS

To derive gene expression prediction models, we used ancestry-stratified datasets from the Multi-Ethnic Study of Atherosclerosis (MESA, including n = 229 African American and n = 381 Hispanic/Latino participants, monocytes) and the Depression Genes and Networks study (DGN, n = 922 European ancestry participants, whole blood). We then performed a transcriptome-wide association study (TWAS) for platelet count, hemoglobin, hematocrit, and white blood cell count in African (n = 27,955) and Hispanic/Latino (n = 28,324) ancestry participants.

RESULTS

Our results revealed 24 suggestive signals (p < 1 × 10-4) that were conditionally distinct from known GWAS identified variants and successfully replicated these signals in European ancestry subjects from UK Biobank. We found modestly improved correlation of predicted and measured gene expression in an independent African American cohort (the Genetic Epidemiology Network of Arteriopathy (GENOA) study (n = 802), lymphoblastoid cell lines) using the larger DGN reference panel; however, some genes were well predicted using MESA but not DGN.

CONCLUSIONS

These analyses demonstrate the importance of performing TWAS and other genetic analyses across diverse populations and of balancing sample size and ancestry background matching when selecting a TWAS reference panel.

Genome-Wide Association of Kidney Traits in Hispanics/Latinos Using Dense Imputed Whole-Genome Sequencing Data: The Hispanic Community Health Study/Study of Latinos

BACKGROUND

Genetic factors that influence kidney traits have been understudied for low-frequency and ancestry-specific variants.

METHODS

This study used imputed whole-genome sequencing from the Trans-Omics for Precision Medicine project to identify novel loci for estimated glomerular filtration rate and urine albumin-to-creatinine ratio in up to 12 207 Hispanics/Latinos. Replication was performed in the Women's Health Initiative and the UK Biobank when variants were available.

RESULTS

Two low-frequency intronic variants were associated with estimated glomerular filtration rate (rs58720902 at AQR, minor allele frequency=0.01, P=1.6×10^-8) or urine albumin-to-creatinine ratio (rs527493184 at ZBTB16, minor allele frequency=0.002, P=1.1×10^-8). An additional variant at PRNT (rs2422935, minor allele frequency=0.54, P=2.89×10^-8) was significantly associated with estimated glomerular filtration rate in meta-analysis with replication samples. We also identified 2 known loci for urine albumin-to-creatinine ratio (BCL2L11 rs116907128, P=5.6×10^-8 and HBB rs344, P=9.3×10^-11) and validated 8 loci for urine albumin-to-creatinine ratio previously identified in the UK Biobank.

CONCLUSIONS

Our study shows gains in gene discovery when using dense imputation from multi-ethnic whole-genome sequencing data in admixed Hispanics/Latinos. It also highlights limitations in genetic research of kidney traits, including the lack of suitable replication samples for variants that are more common in non-European ancestry and those at low frequency in populations.

Allelic Heterogeneity at the CRP Locus Identified by Whole-Genome Sequencing in Multi-ancestry Cohorts

Whole-genome sequencing (WGS) can improve assessment of low-frequency and rare variants, particularly in non-European populations that have been underrepresented in existing genomic studies. The genetic determinants of C-reactive protein (CRP), a biomarker of chronic inflammation, have been extensively studied, with existing genome-wide association studies (GWASs) conducted in >200,000 individuals of European ancestry. In order to discover novel loci associated with CRP levels, we examined a multi-ancestry population (n = 23,279) with WGS (∼38× coverage) from the Trans-Omics for Precision Medicine (TOPMed) program. We found evidence for eight distinct associations at the CRP locus, including two variants that have not been identified previously (rs11265259 and rs181704186), both of which are non-coding and more common in individuals of African ancestry (∼10% and ∼1% minor allele frequency, respectively, and rare or monomorphic in 1000 Genomes populations of East Asian, South Asian, and European ancestry). We show that the minor (G) allele of rs181704186 is associated with lower CRP levels and decreased transcriptional activity and protein binding in vitro, providing a plausible molecular mechanism for this African ancestry-specific signal. The individuals homozygous for rs181704186-G have a mean CRP level of 0.23 mg/L, in contrast to individuals heterozygous for rs181704186 with mean CRP of 2.97 mg/L and major allele homozygotes with mean CRP of 4.11 mg/L. This study demonstrates the utility of WGS in multi-ethnic populations to drive discovery of complex trait associations of large effect and to identify functional alleles in noncoding regulatory regions.

Use of >100,000 NHLBI Trans-Omics for Precision Medicine (TOPMed) Consortium whole genome sequences improves imputation quality and detection of rare variant associations in admixed African and Hispanic/Latino populations

Most genome-wide association and fine-mapping studies to date have been conducted in individuals of European descent, and genetic studies of populations of Hispanic/Latino and African ancestry are limited. In addition, these populations have more complex linkage disequilibrium structure. In order to better define the genetic architecture of these understudied populations, we leveraged >100,000 phased sequences available from deep-coverage whole genome sequencing through the multi-ethnic NHLBI Trans-Omics for Precision Medicine (TOPMed) program to impute genotypes into admixed African and Hispanic/Latino samples with genome-wide genotyping array data. We demonstrated that using TOPMed sequencing data as the imputation reference panel improves genotype imputation quality in these populations, which subsequently enhanced gene-mapping power for complex traits. For rare variants with minor allele frequency (MAF) < 0.5%, we observed a 2.3- to 6.1-fold increase in the number of well-imputed variants, with 11-34% improvement in average imputation quality, compared to the state-of-the-art 1000 Genomes Project Phase 3 and Haplotype Reference Consortium reference panels. Impressively, even for extremely rare variants with minor allele count <10 (including singletons) in the imputation target samples, average information content rescued was >86%. Subsequent association analyses of TOPMed reference panel-imputed genotype data with hematological traits (hemoglobin (HGB), hematocrit (HCT), and white blood cell count (WBC)) in ~21,600 African-ancestry and ~21,700 Hispanic/Latino individuals identified associations with two rare variants in the HBB gene (rs33930165 with higher WBC [p = 8.8x10-15] in African populations, rs11549407 with lower HGB [p = 1.5x10-12] and HCT [p = 8.8x10-10] in Hispanics/Latinos). By comparison, neither variant would have been genome-wide significant if either 1000 Genomes Project Phase 3 or Haplotype Reference Consortium reference panels had been used for imputation. Our findings highlight the utility of the TOPMed imputation reference panel for identification of novel rare variant associations not previously detected in similarly sized genome-wide studies of under-represented African and Hispanic/Latino populations.

Expression of tryptophan metabolizing enzymes (TMEs) and its transporter, LAT1, in metastatic melanoma (MM): Prognostic and therapeutic implications

e21014

Background: The ECHO-301/KEYNOTE-252 trial failed to show clinical benefit of adding epacadostat to pembrolizumab in PD-1 inhibitor-naïve MM. We reasoned that the expression of other TMEs and/or LAT1 by melanoma cells plays a more important role. Methods: Melanoma tissues from stage III/IV pts were stained for the 4 TMEs (TPH1, TPH2, TDO2, IDO1) and LAT1 by single-color immunohistochemistry. Tissues were scored for the status of tumor-infiltrating lymphocytes (TILs) and the expression of 5 proteins separately in melanoma cells and TILs (if present). Association between protein expression, TIL status, and melanoma-specific overall survival was performed. Results: Tissues from 87 pts (stage IV, n = 25) were available. Expression of all 4 TMEs and LAT1 was significantly higher in melanoma cells compared to TILs ( p< 0.001); TPH1, TPH2, and TDO2 expression in melanoma cells was significantly higher than IDO1 (p < 0.001). Lower TPH1 expression in melanoma cells was associated with presence of TILs. Multivariate analysis using a Cox proportional hazards model that included expression of the 5 proteins in melanoma cells showed that high IDO1 was a favorable and high LAT1 expression was an adverse prognostic factor. Conclusions: Expression of TMEs and LAT1 by melanoma cells may have more important role in metastatic melanoma by depleting an essential amino acid from the tumor microenvironment. Pharmacologic targeting of TPH1/TPH2 (e.g. telotristat) may be more clinically relevant in MM as opposed to IDO1 inhibition.