RESUMEN
Genetic alterations in signaling pathways that control cell-cycle progression, apoptosis, and cell growth are common hallmarks of cancer, but the extent, mechanisms, and co-occurrence of alterations in these pathways differ between individual tumors and tumor types. Using mutations, copy-number changes, mRNA expression, gene fusions and DNA methylation in 9,125 tumors profiled by The Cancer Genome Atlas (TCGA), we analyzed the mechanisms and patterns of somatic alterations in ten canonical pathways: cell cycle, Hippo, Myc, Notch, Nrf2, PI-3-Kinase/Akt, RTK-RAS, TGFß signaling, p53 and ß-catenin/Wnt. We charted the detailed landscape of pathway alterations in 33 cancer types, stratified into 64 subtypes, and identified patterns of co-occurrence and mutual exclusivity. Eighty-nine percent of tumors had at least one driver alteration in these pathways, and 57% percent of tumors had at least one alteration potentially targetable by currently available drugs. Thirty percent of tumors had multiple targetable alterations, indicating opportunities for combination therapy.
Asunto(s)
Bases de Datos Genéticas , Neoplasias/patología , Transducción de Señal/genética , Genes Relacionados con las Neoplasias , Humanos , Neoplasias/genética , Fosfatidilinositol 3-Quinasas/genética , Fosfatidilinositol 3-Quinasas/metabolismo , Factor de Crecimiento Transformador beta/genética , Factor de Crecimiento Transformador beta/metabolismo , Proteína p53 Supresora de Tumor/genética , Proteína p53 Supresora de Tumor/metabolismo , Proteínas Wnt/genética , Proteínas Wnt/metabolismoRESUMEN
Evolving in sync with the computation revolution over the past 30 years, computational biology has emerged as a mature scientific field. While the field has made major contributions toward improving scientific knowledge and human health, individual computational biology practitioners at various institutions often languish in career development. As optimistic biologists passionate about the future of our field, we propose solutions for both eager and reluctant individual scientists, institutions, publishers, funding agencies, and educators to fully embrace computational biology. We believe that in order to pave the way for the next generation of discoveries, we need to improve recognition for computational biologists and better align pathways of career success with pathways of scientific progress. With 10 outlined steps, we call on all adjacent fields to move away from the traditional individual, single-discipline investigator research model and embrace multidisciplinary, data-driven, team science.
Asunto(s)
Biología Computacional , Presupuestos , Conducta Cooperativa , Humanos , Investigación Interdisciplinaria , Tutoría , Motivación , Publicaciones , Recompensa , Programas InformáticosRESUMEN
A variational autoencoder (VAE) is a machine learning algorithm, useful for generating a compressed and interpretable latent space. These representations have been generated from various biomedical data types and can be used to produce realistic-looking simulated data. However, standard vanilla VAEs suffer from entangled and uninformative latent spaces, which can be mitigated using other types of VAEs such as ß-VAE and MMD-VAE. In this project, we evaluated the ability of VAEs to learn cell morphology characteristics derived from cell images. We trained and evaluated these three VAE variants-Vanilla VAE, ß-VAE, and MMD-VAE-on cell morphology readouts and explored the generative capacity of each model to predict compound polypharmacology (the interactions of a drug with more than one target) using an approach called latent space arithmetic (LSA). To test the generalizability of the strategy, we also trained these VAEs using gene expression data of the same compound perturbations and found that gene expression provides complementary information. We found that the ß-VAE and MMD-VAE disentangle morphology signals and reveal a more interpretable latent space. We reliably simulated morphology and gene expression readouts from certain compounds thereby predicting cell states perturbed with compounds of known polypharmacology. Inferring cell state for specific drug mechanisms could aid researchers in developing and identifying targeted therapeutics and categorizing off-target effects in the future.
Asunto(s)
Aprendizaje Automático , Polifarmacología , AlgoritmosRESUMEN
BACKGROUND: We have identified molecules that exhibit synthetic lethality in cells with loss of the neurofibromin 1 (NF1) tumor suppressor gene. However, recognizing tumors that have inactivation of the NF1 tumor suppressor function is challenging because the loss may occur via mechanisms that do not involve mutation of the genomic locus. Degradation of the NF1 protein, independent of NF1 mutation status, phenocopies inactivating mutations to drive tumors in human glioma cell lines. NF1 inactivation may alter the transcriptional landscape of a tumor and allow a machine learning classifier to detect which tumors will benefit from synthetic lethal molecules. RESULTS: We developed a strategy to predict tumors with low NF1 activity and hence tumors that may respond to treatments that target cells lacking NF1. Using RNAseq data from The Cancer Genome Atlas (TCGA), we trained an ensemble of 500 logistic regression classifiers that integrates mutation status with whole transcriptomes to predict NF1 inactivation in glioblastoma (GBM). On TCGA data, the classifier detected NF1 mutated tumors (test set area under the receiver operating characteristic curve (AUROC) mean = 0.77, 95% quantile = 0.53 - 0.95) over 50 random initializations. On RNA-Seq data transformed into the space of gene expression microarrays, this method produced a classifier with similar performance (test set AUROC mean = 0.77, 95% quantile = 0.53 - 0.96). We applied our ensemble classifier trained on the transformed TCGA data to a microarray validation set of 12 samples with matched RNA and NF1 protein-level measurements. The classifier's NF1 score was associated with NF1 protein concentration in these samples. CONCLUSIONS: We demonstrate that TCGA can be used to train accurate predictors of NF1 inactivation in GBM. The ensemble classifier performed well for samples with very high or very low NF1 protein concentrations but had mixed performance in samples with intermediate NF1 concentrations. Nevertheless, high-performing and validated predictors have the potential to be paired with targeted therapies and personalized medicine.
Asunto(s)
Biología Computacional/métodos , Silenciador del Gen , Glioblastoma/patología , Aprendizaje Automático , Neurofibromina 1/genética , Transcriptoma , Línea Celular Tumoral , HumanosRESUMEN
In profiling assays, thousands of biological properties are measured in a single test, yielding biological discoveries by capturing the state of a cell population, often at the single-cell level. However, for profiling datasets, it has been challenging to evaluate the phenotypic activity of a sample and the phenotypic consistency among samples, due to profiles' high dimensionality, heterogeneous nature, and non-linear properties. Existing methods leave researchers uncertain where to draw boundaries between meaningful biological response and technical noise. Here, we developed a statistical framework that uses the well-established mean average precision (mAP) as a single, data-driven metric to bridge this gap. We validated the mAP framework against established metrics through simulations and real-world data applications, revealing its ability to capture subtle and meaningful biological differences in cell state. Specifically, we used mAP to assess both phenotypic activity for a given perturbation (or a sample) as well as consistency within groups of perturbations (or samples) across diverse high-dimensional datasets. We evaluated the framework on different profile types (image, protein, and mRNA profiles), perturbation types (CRISPR gene editing, gene overexpression, and small molecules), and profile resolutions (single-cell and bulk). Our open-source software allows this framework to be applied to identify interesting biological phenomena and promising therapeutics from large-scale profiling data.
RESUMEN
BACKGROUND: High-grade serous carcinoma (HGSC) gene expression subtypes are associated with differential survival. We characterized HGSC gene expression in Black individuals and considered whether gene expression differences by self-identified race may contribute to poorer HGSC survival among Black versus White individuals. METHODS: We included newly generated RNA sequencing data from Black and White individuals and array-based genotyping data from four existing studies of White and Japanese individuals. We used K-means clustering, a method with no predefined number of clusters or dataset-specific features, to assign subtypes. Cluster- and dataset-specific gene expression patterns were summarized by moderated t-scores. We compared cluster-specific gene expression patterns across datasets by calculating the correlation between the summarized vectors of moderated t-scores. After mapping to The Cancer Genome Atlas-derived HGSC subtypes, we used Cox proportional hazards models to estimate subtype-specific survival by dataset. RESULTS: Cluster-specific gene expression was similar across gene expression platforms and racial groups. Comparing the Black population with the White and Japanese populations, the immunoreactive subtype was more common (39% vs. 23%-28%) and the differentiated subtype was less common (7% vs. 22%-31%). Patterns of subtype-specific survival were similar between the Black and White populations with RNA sequencing data; compared with mesenchymal cases, the risk of death was similar for proliferative and differentiated cases and suggestively lower for immunoreactive cases [Black population HR = 0.79 (0.55, 1.13); White population HR = 0.86 (0.62, 1.19)]. CONCLUSIONS: Although the prevalence of HGSC subtypes varied by race, subtype-specific survival was similar. IMPACT: HGSC subtypes can be consistently assigned across platforms and self-identified racial groups.
Asunto(s)
Cistadenocarcinoma Seroso , Neoplasias Ováricas , Humanos , Femenino , Neoplasias Ováricas/genética , Neoplasias Ováricas/etnología , Neoplasias Ováricas/patología , Neoplasias Ováricas/mortalidad , Cistadenocarcinoma Seroso/genética , Cistadenocarcinoma Seroso/patología , Cistadenocarcinoma Seroso/etnología , Cistadenocarcinoma Seroso/mortalidad , Persona de Mediana Edad , Población Blanca/genética , Población Blanca/estadística & datos numéricos , Clasificación del Tumor , Anciano , Negro o Afroamericano/genética , Negro o Afroamericano/estadística & datos numéricosRESUMEN
The morphology of cells is dynamic and mediated by genetic and environmental factors. Characterizing how genetic variation impacts cell morphology can provide an important link between disease association and cellular function. Here, we combine genomic sequencing and high-content imaging approaches on iPSCs from 297 unique donors to investigate the relationship between genetic variants and cellular morphology to map what we term cell morphological quantitative trait loci (cmQTLs). We identify novel associations between rare protein altering variants in WASF2, TSPAN15, and PRLR with several morphological traits related to cell shape, nucleic granularity, and mitochondrial distribution. Knockdown of these genes by CRISPRi confirms their role in cell morphology. Analysis of common variants yields one significant association and nominate over 300 variants with suggestive evidence (P < 10-6) of association with one or more morphology traits. We then use these data to make predictions about sample size requirements for increasing discovery in cellular genetic studies. We conclude that, similar to molecular phenotypes, morphological profiling can yield insight about the function of genes and variants.
Asunto(s)
Células Madre Pluripotentes Inducidas , Sitios de Carácter Cuantitativo , Mapeo Cromosómico , Sitios de Carácter Cuantitativo/genética , Núcleo Celular , Forma de la Célula , Proteínas MutantesRESUMEN
The field of high content imaging has steadily evolved and expanded substantially across many industry and academic research institutions since it was first described in the early 1990's. High content imaging refers to the automated acquisition and analysis of microscopic images from a variety of biological sample types. Integration of high content imaging microscopes with multiwell plate handling robotics enables high content imaging to be performed at scale and support medium- to high-throughput screening of pharmacological, genetic and diverse environmental perturbations upon complex biological systems ranging from 2D cell cultures to 3D tissue organoids to small model organisms. In this perspective article the authors provide a collective view on the following key discussion points relevant to the evolution of high content imaging: ⢠Evolution and impact of high content imaging: An academic perspective ⢠Evolution and impact of high content imaging: An industry perspective ⢠Evolution of high content image analysis ⢠Evolution of high content data analysis pipelines towards multiparametric and phenotypic profiling applications ⢠The role of data integration and multiomics ⢠The role and evolution of image data repositories and sharing standards ⢠Future perspective of high content imaging hardware and software.
RESUMEN
Drug resistance is a challenge in anticancer therapy. In many cases, cancers can be resistant to the drug prior to exposure, that is, possess intrinsic drug resistance. However, we lack target-independent methods to anticipate resistance in cancer cell lines or characterize intrinsic drug resistance without a priori knowledge of its cause. We hypothesized that cell morphology could provide an unbiased readout of drug resistance. To test this hypothesis, we used HCT116 cells, a mismatch repair-deficient cancer cell line, to isolate clones that were resistant or sensitive to bortezomib, a well-characterized proteasome inhibitor and anticancer drug to which many cancer cells possess intrinsic resistance. We then expanded these clones and measured high-dimensional single-cell morphology profiles using Cell Painting, a high-content microscopy assay. Our imaging- and computation-based profiling pipeline identified morphological features that differed between resistant and sensitive cells. We used these features to generate a morphological signature of bortezomib resistance. We then employed this morphological signature to analyze a set of HCT116 clones (five resistant and five sensitive) that had not been included in the signature training dataset, and correctly predicted sensitivity to bortezomib in seven cases, in the absence of drug treatment. This signature predicted bortezomib resistance better than resistance to other drugs targeting the ubiquitin-proteasome system, indicating specificity for mechanisms of resistance to bortezomib. Our results establish a proof-of-concept framework for the unbiased analysis of drug resistance using high-content microscopy of cancer cells, in the absence of drug treatment.
Asunto(s)
Antineoplásicos , Microscopía , Bortezomib/farmacología , Ácidos Borónicos/farmacología , Ácidos Borónicos/uso terapéutico , Pirazinas/farmacología , Resistencia a Antineoplásicos , Línea Celular Tumoral , Antineoplásicos/farmacología , Inhibidores de Proteasoma/farmacología , Complejo de la Endopetidasa Proteasomal/metabolismo , ApoptosisRESUMEN
Introduction: High-grade serous carcinoma (HGSC) gene expression subtypes are associated with differential survival. We characterized HGSC gene expression in Black individuals and considered whether gene expression differences by race may contribute to poorer HGSC survival among Black versus non-Hispanic White individuals. Methods: We included newly generated RNA-Seq data from Black and White individuals, and array-based genotyping data from four existing studies of White and Japanese individuals. We assigned subtypes using K-means clustering. Cluster- and dataset-specific gene expression patterns were summarized by moderated t-scores. We compared cluster-specific gene expression patterns across datasets by calculating the correlation between the summarized vectors of moderated t-scores. Following mapping to The Cancer Genome Atlas (TCGA)-derived HGSC subtypes, we used Cox proportional hazards models to estimate subtype-specific survival by dataset. Results: Cluster-specific gene expression was similar across gene expression platforms. Comparing the Black study population to the White and Japanese study populations, the immunoreactive subtype was more common (39% versus 23%-28%) and the differentiated subtype less common (7% versus 22%-31%). Patterns of subtype-specific survival were similar between the Black and White populations with RNA-Seq data; compared to mesenchymal cases, the risk of death was similar for proliferative and differentiated cases and suggestively lower for immunoreactive cases (Black population HR=0.79 [0.55, 1.13], White population HR=0.86 [0.62, 1.19]). Conclusions: A single, platform-agnostic pipeline can be used to assign HGSC gene expression subtypes. While the observed prevalence of HGSC subtypes varied by race, subtype-specific survival was similar.
RESUMEN
A key challenge of the modern genomics era is developing data-driven representations of gene function. Here, we present the first unbiased morphology-based genome-wide perturbation atlas in human cells, containing three genome-scale genotype-phenotype maps comprising >20,000 single-gene CRISPR-Cas9-based knockout experiments in >30 million cells. Our optical pooled cell profiling approach (PERISCOPE) combines a de-stainable high-dimensional phenotyping panel (based on Cell Painting1,2) with optical sequencing of molecular barcodes and a scalable open-source analysis pipeline to facilitate massively parallel screening of pooled perturbation libraries. This approach provides high-dimensional phenotypic profiles of individual cells, while simultaneously enabling interrogation of subcellular processes. Our atlas reconstructs known pathways and protein-protein interaction networks, identifies culture media-specific responses to gene knockout, and clusters thousands of human genes by phenotypic similarity. Using this atlas, we identify the poorly-characterized disease-associated transmembrane protein TMEM251/LYSET as a Golgi-resident protein essential for mannose-6-phosphate-dependent trafficking of lysosomal enzymes, showing the power of these representations. In sum, our atlas and screening technology represent a rich and accessible resource for connecting genes to cellular functions at scale.
RESUMEN
Technological advances in high-throughput microscopy have facilitated the acquisition of cell images at a rapid pace, and data pipelines can now extract and process thousands of image-based features from microscopy images. These features represent valuable single-cell phenotypes that contain information about cell state and biological processes. The use of these features for biological discovery is known as image-based or morphological profiling. However, these raw features need processing before use and image-based profiling lacks scalable and reproducible open-source software. Inconsistent processing across studies makes it difficult to compare datasets and processing steps, further delaying the development of optimal pipelines, methods, and analyses. To address these issues, we present Pycytominer, an open-source software package with a vibrant community that establishes an image-based profiling standard. Pycytominer has a simple, user-friendly Application Programming Interface (API) that implements image-based profiling functions for processing high-dimensional morphological features extracted from microscopy images of cells. Establishing Pycytominer as a standard image-based profiling toolkit ensures consistent data processing pipelines with data provenance, therefore minimizing potential inconsistencies and enabling researchers to confidently derive accurate conclusions and discover novel insights from their data, thus driving progress in our field.
RESUMEN
A primary obstacle in translating genetic associations with disease into therapeutic strategies is elucidating the cellular programs affected by genetic risk variants and effector genes. Here, we introduce LipocyteProfiler, a cardiometabolic-disease-oriented high-content image-based profiling tool that enables evaluation of thousands of morphological and cellular profiles that can be systematically linked to genes and genetic variants relevant to cardiometabolic disease. We show that LipocyteProfiler allows surveillance of diverse cellular programs by generating rich context- and process-specific cellular profiles across hepatocyte and adipocyte cell-state transitions. We use LipocyteProfiler to identify known and novel cellular mechanisms altered by polygenic risk of metabolic disease, including insulin resistance, fat distribution, and the polygenic contribution to lipodystrophy. LipocyteProfiler paves the way for large-scale forward and reverse deep phenotypic profiling in lipocytes and provides a framework for the unbiased identification of causal relationships between genetic variants and cellular programs relevant to human disease.
RESUMEN
Pediatric brain and spinal cancers are collectively the leading disease-related cause of death in children; thus, we urgently need curative therapeutic strategies for these tumors. To accelerate such discoveries, the Children's Brain Tumor Network (CBTN) and Pacific Pediatric Neuro-Oncology Consortium (PNOC) created a systematic process for tumor biobanking, model generation, and sequencing with immediate access to harmonized data. We leverage these data to establish OpenPBTA, an open collaborative project with over 40 scalable analysis modules that genomically characterize 1,074 pediatric brain tumors. Transcriptomic classification reveals universal TP53 dysregulation in mismatch repair-deficient hypermutant high-grade gliomas and TP53 loss as a significant marker for poor overall survival in ependymomas and H3 K28-mutant diffuse midline gliomas. Already being actively applied to other pediatric cancers and PNOC molecular tumor board decision-making, OpenPBTA is an invaluable resource to the pediatric oncology community.
RESUMEN
Software has provided cell biologists the power to quantify specific cellular features in cell images at scale. Before long, these biologists also recognized the potential to extract much more biological information from the same images. From here, the field of image-based profiling, the process of extracting unbiased representations that capture morphological cell state, was born. We are still in the early days of image-based profiling, and it is clear that the many opportunities to interrogate biological systems come with significant challenges. These challenges include building expressive and biologically-relevant representations, adjusting for technical noise, writing generalizable software infrastructure, continuing to foster a culture of open science, and promoting FAIR (findable, accessible, interoperable, and reusable) data. We present a workshop at the Pacific Symposium on Biocomputing 2022 to introduce the field of image-based profiling to the broader computational biology community. In the following document, we introduce image-based profiling, discuss current state-of-the-art methods and limitations, and provide rationale for why now is the perfect time for the field to expand. We also introduce our invited speakers and agenda, which together provide an introduction to the field complemented by in-depth application areas in industry and academia. We also include five lightning talks to complement the invited speakers on various methodological and discovery advances.
Asunto(s)
Biología Computacional , Programas InformáticosRESUMEN
Morphological and gene expression profiling can cost-effectively capture thousands of features in thousands of samples across perturbations by disease, mutation, or drug treatments, but it is unclear to what extent the two modalities capture overlapping versus complementary information. Here, using both the L1000 and Cell Painting assays to profile gene expression and cell morphology, respectively, we perturb human A549 lung cancer cells with 1,327 small molecules from the Drug Repurposing Hub across six doses, providing a data resource including dose-response data from both assays. The two assays capture both shared and complementary information for mapping cell state. Cell Painting profiles from compound perturbations are more reproducible and show more diversity but measure fewer distinct groups of features. Applying unsupervised and supervised methods to predict compound mechanisms of action (MOAs) and gene targets, we find that the two assays not only provide a partially shared but also a complementary view of drug mechanisms. Given the numerous applications of profiling in biology, our analyses provide guidance for planning experiments that profile cells for detecting distinct cell types, disease phenotypes, and response to chemical or genetic perturbations.
Asunto(s)
Perfilación de la Expresión Génica , Humanos , Perfilación de la Expresión Génica/métodos , FenotipoRESUMEN
Genetic and chemical perturbations impact diverse cellular phenotypes, including multiple indicators of cell health. These readouts reveal toxicity and antitumorigenic effects relevant to drug discovery and personalized medicine. We developed two customized microscopy assays, one using four targeted reagents and the other three targeted reagents, to collectively measure 70 specific cell health phenotypes including proliferation, apoptosis, reactive oxygen species, DNA damage, and cell cycle stage. We then tested an approach to predict multiple cell health phenotypes using Cell Painting, an inexpensive and scalable image-based morphology assay. In matched CRISPR perturbations of three cancer cell lines, we collected both Cell Painting and cell health data. We found that simple machine learning algorithms can predict many cell health readouts directly from Cell Painting images, at less than half the cost. We hypothesized that these models can be applied to accurately predict cell health assay outcomes for any future or existing Cell Painting dataset. For Cell Painting images from a set of 1500+ compound perturbations across multiple doses, we validated predictions by orthogonal assay readouts. We provide a web app to browse predictions: http://broad.io/cell-health-app. Our approach can be used to add cell health annotations to Cell Painting datasets.
Asunto(s)
Células/citología , Predicción/métodos , Procesamiento de Imagen Asistido por Computador/métodos , Algoritmos , Bioensayo , Línea Celular , Humanos , Aprendizaje Automático , Microscopía , FenotipoRESUMEN
BACKGROUND: Bone accrual impacts lifelong skeletal health, but genetic discovery has been primarily limited to cross-sectional study designs and hampered by uncertainty about target effector genes. Here, we capture this dynamic phenotype by modeling longitudinal bone accrual across 11,000 bone scans in a cohort of healthy children and adolescents, followed by genome-wide association studies (GWAS) and variant-to-gene mapping with functional follow-up. RESULTS: We identify 40 loci, 35 not previously reported, with various degrees of supportive evidence, half residing in topological associated domains harboring known bone genes. Of several loci potentially associated with later-life fracture risk, a candidate SNP lookup provides the most compelling evidence for rs11195210 (SMC3). Variant-to-gene mapping combining ATAC-seq to assay open chromatin with high-resolution promoter-focused Capture C identifies contacts between GWAS loci and nearby gene promoters. siRNA knockdown of gene expression supports the putative effector gene at three specific loci in two osteoblast cell models. Finally, using CRISPR-Cas9 genome editing, we confirm that the immediate genomic region harboring the putative causal SNP influences PRPF38A expression, a location which is predicted to coincide with a set of binding sites for relevant transcription factors. CONCLUSIONS: Using a new longitudinal approach, we expand the number of genetic loci putatively associated with pediatric bone gain. Functional follow-up in appropriate cell models finds novel candidate genes impacting bone accrual. Our data also raise the possibility that the cell fate decision between osteogenic and adipogenic lineages is important in normal bone accrual.
Asunto(s)
Desarrollo Óseo/genética , Enfermedades Óseas/genética , Huesos , Predisposición Genética a la Enfermedad/genética , Estudio de Asociación del Genoma Completo , Adolescente , Densidad Ósea , Niño , Preescolar , Cromatina , Mapeo Cromosómico , Estudios Transversales , Femenino , Edición Génica , Expresión Génica , Genómica , Humanos , Masculino , Proteínas Nucleares/genética , Proteínas Nucleares/metabolismo , Osteoblastos , Osteogénesis/genética , Fenotipo , Polimorfismo de Nucleótido Simple , Regiones Promotoras Genéticas , Sitios de Carácter Cuantitativo , Proteínas de Unión al ARN/genética , Proteínas de Unión al ARN/metabolismo , Adulto JovenRESUMEN
BACKGROUND: Unsupervised compression algorithms applied to gene expression data extract latent or hidden signals representing technical and biological sources of variation. However, these algorithms require a user to select a biologically appropriate latent space dimensionality. In practice, most researchers fit a single algorithm and latent dimensionality. We sought to determine the extent by which selecting only one fit limits the biological features captured in the latent representations and, consequently, limits what can be discovered with subsequent analyses. RESULTS: We compress gene expression data from three large datasets consisting of adult normal tissue, adult cancer tissue, and pediatric cancer tissue. We train many different models across a large range of latent space dimensionalities and observe various performance differences. We identify more curated pathway gene sets significantly associated with individual dimensions in denoising autoencoder and variational autoencoder models trained using an intermediate number of latent dimensionalities. Combining compressed features across algorithms and dimensionalities captures the most pathway-associated representations. When trained with different latent dimensionalities, models learn strongly associated and generalizable biological representations including sex, neuroblastoma MYCN amplification, and cell types. Stronger signals, such as tumor type, are best captured in models trained at lower dimensionalities, while more subtle signals such as pathway activity are best identified in models trained with more latent dimensionalities. CONCLUSIONS: There is no single best latent dimensionality or compression algorithm for analyzing gene expression data. Instead, using features derived from different compression models across multiple latent space dimensionalities enhances biological representations.