Biology subjects

Fonseca, G.

Publications and source records attributed to Fonseca, G..

7 recordsLinked to original sources

bioRxiv2025

DTractor enhances cell type deconvolution in spatial transcriptomics by integrating deep neural networks, transfer learning, and matrix factorization

Spatial transcriptomics (ST) captures gene expression with spatial context but lacks single-cell resolution. Single-cell RNA sequencing (scRNA-seq) offers high-resolution profiles without spatial information. Accurate spot-level decomposition requires effective integration of both. We present DTractor, a deep learning-based framework that improves cell-type deconvolution in ST data through spatial constraints and transfer learning. DTractor achieves dual utilization of scRNA-seq reference data by incorporating both a cell-type-specific gene expression matrix and learned latent embeddings into a unified matrix factorization model. This joint modeling enables accurate estimation of cell-type proportions and cell-type-resolved gene expression within each spatial spot, while preserving biological and spatial coherence. DTractor further applies spatial regularization to maintain local tissue structure. Across multiple ST platforms and tissue types, DTractor demonstrates improved decomposition accuracy, robustness, and interpretability compared to existing methods. The results from DTractor support downstream applications such as spatial domain analysis and the study of spatially organized cellular behaviors.

bioinformatics↗

bioRxiv2025

Microbial signatures define the ecosystem functions of the pelagic microbiome in a basin-scale, Southwest Atlantic Ocean

BackgroundThe pelagic environment may present a mosaic of biogeographical domains that regional oceanographic processes can influence. Here, a coastal-to-open ocean microbiome investigation was conducted on 64 water samples from the Santos Basin (SB), South Atlantic Ocean. Using metagenomics and machine learning approaches, we assessed the diversity and distribution of pelagic microbes, identified key bacterial and archaeal taxa, and inferred their ecosystem functions. ResultsUnsupervised machine learning revealed a clear spatial and vertical (light availability) distribution pattern across SB, with some indicator taxa previously observed in other marine waters. Supervised learning further revealed that environmental variables, notably phosphate, salinity, and nitrate, which are key markers of local upwelling and the La Plata River plume, are primary drivers of microbial community structure. Furthermore, we recovered 307 metagenome-assembled genomes with 45% of Archaea and 42% of Bacteria possible new taxa. In terms of functionality, the SB dataset revealed a pelagic ecosystem resembling typical marine (e.g., Atlantic Ocean) waters, with photoautotrophs and nitrogen fixers in the photic zone and different autotrophic pathways in the aphotic environment. Surprisingly, the SB dataset revealed genes for CO bio-oxidation and algal dimethylsulfoniopropionate (DMSP) degradation at all depths. Furthermore, we observed potential non- cyanobacterial diazotrophs in dark water. ConclusionsOur results revealed that the SB represents a unique ecosystem with local oceanographic processes shaping the distribution of diverse and uncharacterized microbiomes. Additionally, these findings highlight the importance of mixotrophic microbes in SB biogeochemical cycles. This massive investigation of the SB pelagic microbiome provided knowledge-based data for understanding local ecosystem health, services, and dynamics, which are essential for future sustainable ocean management.

microbiology↗

bioRxiv2025

A Machine Learning Approach Elucidates Spatial Patterns of Environmental Properties Driving Microbial Composition Over Santos Basin, South Atlantic

Microbial communities in marine ecosystems play a pivotal role in global biogeochemical cycles, with particular relevance in ecologically and industrially significant regions such as the Santos Basin (SB), Brazils largest marine sedimentary basin and a hub for oil and gas exploration. Yet, our capacity of predicting the microbial community structure and function remains limited for marine ecosystems. This study investigated the structure of microbial communities across different depths in the SB, using amplicon sequencing data and quantitative cell counts obtained via flow cytometry. Using a hybrid machine learning framework combining Self-Organizing Maps and Random Forest, we identified five distinct microbial assemblages (named microbial associations) in the SB predicted with 86% accuracy. These associations were primarily driven by temperature, water density, salinity, and nutrients such as phosphate and nitrate. Our findings showed a clear stratification of microbial communities across pelagic zones, with temperature as the main factor driving the structure in epipelagic and mesopelagic layers, while salinity and density exerted greater influence in the deeper bathypelagic communities. Temporal and spatial variations, particularly between 2019 and 2021, in regions influenced by the Cabo Frio upwelling and Rio de la Plata plume, further highlighted the impact of regional and local oceanographic processes on community dynamics. The associations from deeper waters harbored more diverse microbial assemblages, and shallow waters, on the other hand, possess higher absolute abundance of microbial cells, suggesting niche specialization across depths. This study underscores the importance of environmental gradients as well as local oceanographic processes in shaping microbial diversity, providing valuable insights into the ecological dynamics of the SB, which are essential for understanding the potential impacts of human activities, such as oil and gas exploration and production, on these critical marine ecosystems.

microbiology↗

bioRxiv2023

RAMEN identifies effective indicators for severe COVID and Long COVID patients

The COVID-19 pandemic has significantly altered global socioeconomic structures and individual lives. Understanding the disease mechanisms and facilitating diagnosis requires comprehending the complex interplay among clinical factors like demographics, symptoms, comorbidities, treatments, lab results, complications, and other metrics, and their relation to outcomes such as disease severity and long term outcomes (e.g., post-COVID-19 condition/long COVID). Conventional correlational methods struggle with indirect and directional connections among these factors, while standard graphical methods like Bayesian networks are computationally demanding for extensive clinical variables. In response, we introduced RAMEN, a methodology that integrates Genetic Algorithms with random walks for efficient Bayesian network inference, designed to map the intricate relationships among clinical variables. Applying RAMEN to the Biobanque quebecoise de la COVID-19 (BQC19) dataset, we identified critical markers for long COVID and varying disease severity. The Bayesian Network, corroborated by existing literature and supported through multi-omics analyses, highlights significant clinical variables linked to COVID-19 outcomes. RAMENs ability to accurately map these connections contributes substantially to developing early and effective diagnostics for severe COVID-19 and long COVID.

bioinformatics↗

bioRxiv2022

The Global Environment Facility approach for allocating biodiversity funding to countries

Biodiversity is not evenly distributed across the globe and some areas have greater potential to contribute to biodiversity conservation than others. Whilst there are multiple ways to determine priority areas for conservation, for a global institution like the Global Environment Facility (GEF), the funding mechanism for the Convention on Biological Diversity and the largest multilateral source of funding for developing countries focused on enhancing biodiversity outcomes and promoting sustainable use, it is important to fund the top-ranked countries whilst also ensuring that all eligible countries are able to undertake some biodiversity conservation actions in accordance with the Convention. To this end, the GEF uses the System for Transparent Allocation of Resources (STAR) to allocate funding in separate funding rounds to eligible countries. This country focus means that all prioritization analyses need to be undertaken within that political framework, while also considering the intrinsic patterns in biodiversity that dont respect national borders. We present the 2018 update of the biodiversity component of GEF-STAR, investigate how the weighting system affects the ranking of countries. We show that top ranked and bottom ranked countries are robust to changes in the weighting of analytical elements, but the weighting can significantly alter the importance of middle ranking countries, affecting their funding allocation. This analysis has been used by the GEF, along with other data, to allocate over $1 billion in biodiversity funding (GEF-7 = $1.2 billion) to improve country and global prospects for conservation. However, this large funding allocation for conservation needs to be set against the vastly larger funding flows that decrease natural values around the world, and the need for systems level change remains evident across the entire planet.

ecology↗

bioRxiv2022

The epigenetic modifier DOT1L regulates gene regulatory networks necessary for cardiac patterning and cardiomyocyte cell cycle withdrawal

Mechanisms by which specific histone modifications regulate distinct gene regulatory networks remain little understood. We investigated how H3K79me2, a modification catalyzed by DOT1L and previously considered a general transcriptional activation mark, regulates gene expression in mammalian cardiogenesis. Early embryonic cardiomyocyte ablation of Dot1l revealed that H3K79me2 does not act as a general transcriptional activator, but rather regulates highly specific gene regulatory networks at two critical cardiogenic junctures: left ventricle patterning and postnatal cardiomyocyte cell cycle withdrawal. Mechanistic analyses revealed that H3K79me2 in two distinct domains, gene bodies and regulatory elements, synergized to promote expression of genes activated by DOT1L. Surprisingly, these analyses also revealed that H3K79me2 in specific regulatory elements contributed to silencing genes usually not expressed in cardiomyocytes. As DOT1L mutants had increased numbers of postnatal mononuclear cardiomyocytes and prolonged cardiomyocyte cell cycle activity, controlled inhibition of DOT1L might be a strategy to promote cardiac regeneration post-injury.

developmental biology↗

bioRxiv2020

Discovering biomarkers of chemotherapy resistance using in vitro evolution in haploid human cells

BackgroundIn vitro evolution and whole genome analysis has proven to be a powerful method for studying the mechanism of action of small molecules in many haploid microbes but has generally not been applied to human cell lines in part because their diploid state complicates the identification of variants that confer drug resistance. To determine if haploid human cell could be used in MOA studies, we evolved resistance to five different anticancer drugs (doxorubicin, gemcitabine, etoposide, topotecan, and paclitaxel) using a near-haploid cell line (HAP1) and then analyzed the genomes of the drug resistant clones, developing a bioinformatic pipeline that involved filtering for high frequency alleles predicted to change protein sequence, or alleles which appeared in the same gene for multiple independent selections with the same compound. Applying the filter to sequences from 28 drug resistant clones identified a set of 21 genes which was strongly enriched for known resistance genes or known drug targets (TOP1, TOP2A, DCK, WDR33, SLCO3A1). In addition, some lines carried structural variants that encompassed additional known resistance genes (ABCB1, WWOX and RRM1). Gene expression knockdown and knockout experiments of 10 validation targets showed a high degree of specificity and accuracy in our calls and demonstrates that the same drug resistance mechanisms found in diverse clinical samples can be evolved, discovered and studied in an isogenic background.

cancer biology↗