Biology subjects

Ong, G.

Publications and source records attributed to Ong, G..

7 recordsLinked to original sources

bioRxiv2025

Breast cancer through the lens of whole transcriptome spatial imaging

Using the worlds first transcriptome-scale spatial imaging of a breast tumor, we assessed what could be learned about one patients disease. We cataloged heterogeneity across 3 morphological regions, 9 spatial domains, 37 cell types, and 1692 pathways. Then, employing a new algorithm for spatially stratified differential expression, we tested >2 million hypotheses about cell types behavior across space. We measured how CD8+ T cells change upon entering the tumor, how cancer cells adapt to nutrient-poor microenvironments, how tumor glycolysis impacts nearby healthy cells, and how low-proliferation zones of the tumor are distinct. We found several instances of druggable biology: the tumors heterogeneity and invasion programs suggested aggressiveness independent of traditional grading; PDCD1 expression and exhaustion limited T-cell activity; cancer cells engaged angiogenic signaling but struggled to proliferate in hypoxic regions; and a tumor subcluster up-regulated lipid metabolism. These findings demonstrate the insights obtainable through spatial profiling of individual tumors.

cancer biology↗

bioRxiv2025

Unified Genomic and Chemical Representations Enable Bidirectional Bio-synthetic Gene Cluster and Natural Product Retrieval

Natural product discovery is increasingly driven by the ability to analyze microbial genomes for biosynthetic gene clusters (BGCs) that encode secondary metabolites. While existing approaches have successfully linked BGCs to broad classes of chemical products, they typically operate in a single modality (genomic or chemical) limiting the scope of bidirectional prediction. In this work, we propose a multimodal framework that integrates genomic and chemical information by projecting embeddings derived from pretrained language models into a common representation space. We embed genomic sequences using a BGC foundation model and represent molecules through a chemical language model, then use a metric learning model to co-embed BGCs and their associated chemical structures. This co-embedding space allows us to quantify the similarity between BGCs and compounds using similarity measures, enabling both forward and inverse retrieval tasks. Beyond retrieval, we show that the shared space can guide strain selection for targeted compounds. By identifying BGCs closest to a query compound in the embedding space, we prioritize microbial strains that encode similar clusters, thereby streamlining genome mining and retrobiosynthetic design efforts. This approach represents a generalizable, scalable strategy to bridge biological and chemical modalities in natural product discovery.

synthetic biology↗

bioRxiv2024

Paradigms of convergent evolution in enzymes

There are many occurrences of enzymes catalysing the same reaction but having significantly different structures. Leveraging the comprehensive information on enzymes stored in the Mechanism and Catalytic Site Atlas (M-CSA), we present a collection of 38 cases for which there is sufficient evidence of functional convergence without an evolutionary link. For each case, we compare enzymes which have identical Enzyme Commission numbers (i.e. catalyse the same reaction), but different identifiers in the CATH data resource (i.e. different folds). We focus on similarities between their sequence, structure, active site geometry, cofactors and catalytic mechanism. These features are then assessed to evaluate whether all the evidence on these structurally diverse proteins supports their independent evolution to catalyse the same chemical reaction. Our approach combines literature information with knowledge-based computational resources from, amongst others, M-CSA, PDBe and PDBsum, supported by tailor made software to explore active site structure and assess mechanism similarity. We find that there are multiple varieties of convergent functional evolution observed to date and it is necessary to investigate sequence, structure, active site geometry and enzyme mechanisms to describe such convergence accurately.

evolutionary biology↗

bioRxiv2023

A Machine Learning One-Class Logistic Regression Model to Predict Stemness in Single Cell Transcriptomics and Spatial Omics Datasets

Cell annotation is a crucial methodological component to interpreting single cell and spatial omics data. These approaches are often biased and manually curated. Here we harness an existing stemness model for assessing oncogenic states to transform its application to single cell and spatial omic datasets. This one-class logistic regression machine learning algorithm is used to extract transcriptomic or proteomic features from non-transformed stem cells to identify dedifferentiated cell states. We found this method identifies single cell states in metastatic tumor cell populations without the requirement of cell annotation. Finally these stemness indices are applicable across a variety of spatial transcriptomic and proteomic technologies for the identification of oncogenic cell types in the tumor microenvironment. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=162 SRC="FIGDIR/small/539461v1_ufig1.gif" ALT="Figure 1"> View larger version (37K): org.highwire.dtl.DTLVardef@f2973forg.highwire.dtl.DTLVardef@a7aa9corg.highwire.dtl.DTLVardef@1b1ea27org.highwire.dtl.DTLVardef@183b4a5_HPS_FORMAT_FIGEXP M_FIG C_FIG

genomics↗

bioRxiv2022

Ultra High-Plex Spatial Proteogenomic Investigation of Giant Cell Glioblastoma Multiforme Immune Infiltrates Reveals Distinct Protein and RNA Expression Profiles

A deeper understanding of complex biological processes, including tumor development and immune response, requires ultra high-plex, spatial interrogation of multiple "omes". Here we present the development and implementation of a novel spatial proteogenomic (SPG) assay on the GeoMx(R) Digital Spatial Profiler platform with NGS readout that enables ultra high-plex digital quantitation of proteins (> 100-plex) and RNA (whole transcriptome, > 18,000-plex) from a single FFPE sample. This study highlighted the high concordance, R > 0.85, and <11% change in sensitivity between SPG assay and the single analyte -assays on various cell lines and tissues from human and mouse. Furthermore, we demonstrate that the SPG assay was reproducible across multiple users. When used in conjunction with advanced cellular neighborhood segmentation, distinct immune or tumor RNA and protein targets were spatially resolved within individual cell subpopulations in human colorectal cancer and non-small cell lung cancer. We used the SPG assay to interrogate 23 different glioblastoma multiforme samples across 4 pathologies. The study revealed distinct clustering of both RNA and protein based on pathology and anatomic location. The in-depth investigation of giant cell glioblastoma multiforme revealed distinct protein and RNA expression profiles compared to that of the more common glioblastoma multiforme. More importantly, the use of spatial proteogenomics allowed simultaneous interrogation of critical protein post-translational modifications alongside whole transcriptomic profiles within the same distinct cellular neighborhoods.

cancer biology↗

bioRxiv2021

High-Plex Multiomic Analysis in FFPE Tissue at Single-Cellular and Subcellular Resolution by Spatial Molecular Imaging

The Spatial Molecular Imaging platform (CosMxTM SMI, NanoString Technologies, Seattle, WA) utilizes high-plex in-situ imaging chemistry for both RNA and protein detection. This automated instrument provides 1000s of plex, at high sensitivity (1 to 2 copies/cell), very low error rate (0.0092 false calls/cell) and background ([~]0.04 counts/cell). The imaging system generates three-dimensional super-resolution localization of analytes at [~]2 million cells per sample, four samples per run. Cell segmentation is morphology-based using antibodies, compatible with FFPE samples. Multiomic data (980 RNAs, 108 proteins) were measured at subcellular resolution using FFPE tissues (non-small cell lung (NSCLC) and breast cancer) and allowed identification of over 18 distinct cell types, 10 unique tumor microenvironments, and 100 pairwise ligand-receptor interactions. Over 800,000 single cells and [~]260 million transcripts data are released into the public domain allowing extended data analysis by the entire spatial biology research community.

genomics↗

bioRxiv2018

A polyploid admixed origin of beer yeasts derived from European and Asian wine populations

Strains of Saccharomyces cerevisiae used to make beer, bread and wine are genetically and phenotypically distinct from wild populations associated with trees. The origins of these domesticated populations are not always clear; human-associated migration and admixture with wild populations have had a strong impact on S. cerevisiae population structure. We examined the population genetic history of beer strains and find that ale strains and the S. cerevisiae portion of allotetraploid lager strains were derived from admixture between populations closely related to European grape wine strains and Asian rice wine strains. Similar to both lager and baking strains, ale strains are polyploid, providing them with a passive means of remaining isolated from other populations and providing us with a living relic of their ancestral hybridization. To reconstruct their polyploid origin we phased the genomes of two ale strains and found ale haplotypes to both be recombinants between European and Asian alleles and to also contain novel alleles derived from extinct or as yet uncharacterized populations. We conclude that modern beer strains are the product of a historical melting pot of fermentation technology.

evolutionary biology↗