Biology subjects

Polo, J.

Publications and source records attributed to Polo, J..

5 recordsLinked to original sources

bioRxiv2026

Graph-based RNA structural representation reveals determinants of subcellular localization

RNA subcellular localization is a key determinant of RNA function and regulation, yet existing computational approaches rely primarily on sequence or simplified structural descriptors, limiting their scalability to long transcripts, their ability to model inter-label dependencies, and their applicability across RNA types. Here, we present GRASP, a unified graph neural network framework for predicting RNA subcellular localization using a heterogeneous graph representation that is RNA substructure-aware. GRASP presents each RNA as a multi-scale graph comprising nucleotide nodes and secondary-structure-derived substructure nodes, connected by relational edges, enabling joint modeling of base-level interactions and regional structural context. The model further incorporates multi-label dependency learning to capture co-localization patterns across cellular compartments within a unified framework. Across multiple benchmark datasets and RNA types, GRASP consistently outperforms state-of-the-art sequence-based and structure-informed methods, achieving substantial improvements in accuracy, F1 score, and AUC while maintaining strong scalability to long transcripts. In addition, the graph-based representation provides biologically interpretable insights into structural determinants of RNA localization. The source code and data are available at https://github.com/ABILiLab/GRASP, and the web server is accessible at http://grasp.biotools.bio.

bioinformatics↗

bioRxiv2025

LeGO-Teknik reveals that the neurogenesis pathway is a clone-specific hallmark of brain metastasis in breast cancer

Breast cancer exhibits substantial inter- and intra-patient heterogeneity. Yet the molecular features underlying this diversity and their roles in tumour progression are not yet fully elucidated. Within a primary tumour, certain clones possess a unique capacity to metastasize to distant organs, such as the brain, and this propensity may be associated with specific gene expression profiles. To investigate this phenomenon, we developed a new optical barcoding library, LeGO-Teknik, compatible with high resolution imaging, flow cytometry, single-cell RNA sequencing and machine learning computational methods. Using this new tool, we systematically assessed the fitness and organ tropism of multiple human breast cancer clones in spontaneous metastasis assays. Linking transcriptomic profiles of individual cancer cells to their clonal identities and metastatic behaviours, we identified that clones predisposed to forming brain metastases exhibit distinct molecular signatures. Notably, individual cells from these clones showed enriched expression of neurogenesis-related genes. Using SCENIC transcription factor analysis, we found that some of these features are driven by the transcription factor SOX4. This hallmark, likely due to a process of cellular reprogramming, is a remarkable example of organ mimicry during clonal selection. Together, this work provides critical insights into brain metastasis and highlights the potential for identifying predictive biomarkers and therapeutic targets for metastatic breast cancer.

cancer biology↗

bioRxiv2024

STAMP: Single-Cell Transcriptomics Analysis and Multimodal Profiling through Imaging

We introduce Single-Cell Transcriptomics Analysis and Multimodal Profiling (STAMP), a scalable profiling approach of individual cells. Leveraging transcriptomics and proteomics imaging platforms, STAMP eliminates sequencing costs, to enable single-cell genomics from hundreds to millions of cells at an unprecedented low cost. Stamping cells in suspension onto imaging slides, STAMP supports single-modal (RNA or protein) and multimodal (RNA and protein) profiling and flexible, ultra-high-throughput formats. STAMP allows the analysis of a single or multiple samples within the same experiment, enhancing experimental flexibility, throughput and scale. We tested STAMP with diverse sample types, including peripheral blood mononuclear cells (PBMCs), dissociated cancer cells and differentiated embryonic stem cell cultures, as well as whole cells and nuclei. Combining RNA and protein profiling, we applied immuno-phenotyping of millions of blood cells simultaneously. We also used STAMP to identify ultra-rare cell populations, simulating clinical applications to identify circulating tumor cells (CTCs). Performing in vitro differentiation studies, we further showed its potential for large-scale perturbation studies. Together, STAMP establishes a new standard for cost-effective, scalable single-cell analysis. Without the need for sequencing, STAMP makes high-resolution profiling more affordable and accessible. Designed to meet the needs of research labs, diagnostic cores and pharmaceutical companies, STAMP holds the promise to transform our capacity to map human biology, diagnose diseases and drug discovery.

genomics↗

bioRxiv2022

Parallel use of pluripotent human stem cell lung and heart models provide new insights for treatment of SARS-CoV-2

SARS-CoV-2 primarily infects the respiratory tract, but pulmonary and cardiac complications occur in severe COVID-19. To elucidate molecular mechanisms in the lung and heart, we conducted paired experiments in human stem cell-derived lung alveolar type II (AT2) epithelial cell and cardiac cultures infected with SARS-CoV-2. With CRISPR- Cas9 mediated knock-out of ACE2, we demonstrated that angiotensin converting enzyme 2 (ACE2) was essential for SARS-CoV-2 infection of both cell types but further processing in lung cells required TMPRSS2 while cardiac cells required the endosomal pathway. Host responses were significantly different; transcriptome profiling and phosphoproteomics responses depended strongly on the cell type. We identified several antiviral compounds with distinct antiviral and toxicity profiles in lung AT2 and cardiac cells, highlighting the importance of using several relevant cell types for evaluation of antiviral drugs. Our data provide new insights into rational drug combinations for effective treatment of a virus that affects multiple organ systems. One-sentence summaryRational treatment strategies for SARS-CoV-2 derived from human PSC models

microbiology↗

bioRxiv2022

snPATHO-seq: unlocking the FFPE archives for single nucleus RNA profiling

FFPE (formalin-fixed, paraffin-embedded) tissue archives are the largest repository of clinically annotated human specimens. Despite numerous advances in technology, current methods for sequencing of FFPE-fixed single-cells are slow, labour intensive, insufficiently sensitive and have a low resolution, making it difficult to fully exploit their enormous research and clinical potential. Here we introduce single nuclei pathology sequencing (snPATHO-Seq), a sensitive and efficient high-throughput platform to profile the transcriptome of single nuclei extracted from formalin-fixed paraffin-embedded (FFPE) samples. snPATHO-Seq combines an optimised nuclei extraction protocol from archival samples with 10x Genomics probe-based technology targeting the whole transcriptome. We performed direct comparison of the Fixed RNA Profiling (FRP) and established 3 single cell RNA-Sequencing (scRNA-Seq) workflows through a comprehensive bioinformatics analysis of matched fresh and fixed samples derived from the LNCaP prostate cancer cell line. FRP detected 2.1 times more transcripts in the fixed sample than the 3 kit did in the fresh sample. Low mitochondrial genes detection using the FRP was translated into 99.9 percent of cells passing the QC filters, compared to 81.6 percent of cells using the v3.1 chemistry. We then optimized snPATHO-Seq and applied it to a human breast cancer metastasis to the liver collected at autopsy and preserved in FFPE, a particularly challenging sample type. Remarkably, at 28,000 reads/cell snPATHO-Seq was able to detect a median of 1850 genes/cell and 3,216 UMI counts/cell. Comparison of snPATHO-Seq with spatial transcriptomics data (10x Genomics Visium FFPE v1) derived from an adjacent section of the same sample revealed a strong correlation, validating the accuracy of the snPATHO-Seq data. Gene expression data from snPATHO-Seq was used to predict cell type composition within each spatial transcriptomic location via deconvolution. Overall, snPATHO-Seq enables high quality and sensitivity snRNA-Seq from preserved tissue samples, unlocking the vast archives of FFPE tissues and thereby allowing extensive retrospective clinical genomic studies.

cancer biology↗