Biology subjects

Brown, D. V.

Publications and source records attributed to Brown, D. V..

7 recordsLinked to original sources

bioRxiv2026

Active Learning Enables Efficient Directed Evolution of a Far-Red Fluorescent Protein with Minimal Experimental Data

Fluorescent proteins are fundamental tools for cellular imaging. Most fluorescent proteins in routine use, including GFP, are derived from the jellyfish Aequorea victoria and emit blue-green light, which is strongly absorbed and scattered by tissue, limiting imaging depth. Far-red and near-infrared fluorescent proteins, engineered from bacteriophytochromes, address this limitation because far-red light penetrates tissue considerably further. However, these proteins are typically much dimmer than their A. victoria -derived counterparts. Improving brightness by conventional directed evolution requires screening large random mutant libraries, a process that is slow, labor-intensive, and often impractical outside specialized laboratories. We utilized an active-learning-guided directed evolution workflow that identified improved variants from substantially less data than conventional screening. Each round coupled automated, miniaturized cell-free protein expression directly from a DNA template without cloning or cell culture, with a machine-learning model retrained on cumulative sequence-function data to nominate the most informative variants for the next round. Applied to miRFP670nano3, this workflow screened 120 variants across successive rounds and identified twelve with improved brightness, the best four-fold brighter in bacterial systems. However, these gains did not translate when the variants were evaluated in mammalian cells, indicating that performance can be strongly dependent on cellular context. Retrospective simulation across benchmark datasets from ProteinGym showed that performing more experimental batches with fewer samples per batch consistently accelerated convergence to high-fitness sequences. Incorporating protein-language-model derived zero-shot fitness priors also accelerated convergence, but only in proportion to how well each prior score correlated with the true fitness landscape. Together, these findings established generalizable design rules, favoring smaller acquisition batches and confidence-weighted priors, for engineering proteins from minimal experimental data. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=55 SRC="FIGDIR/small/744534v1_ufig1.gif" ALT="Figure 1"> View larger version (11K): org.highwire.dtl.DTLVardef@14cd238org.highwire.dtl.DTLVardef@7d8334org.highwire.dtl.DTLVardef@30e9f2org.highwire.dtl.DTLVardef@14f3a0c_HPS_FORMAT_FIGEXP M_FIG C_FIG

synthetic biology↗

bioRxiv2025

CMDdemux: an efficient single cell demultiplexing method

Multiplexing technologies label cells with molecular tags, allowing cells from different donors to be pooled together for sequencing. Although this approach enhances cell throughput, eliminates batch effects, and enables doublet detection, limitations of hashtag-based labelling can still lead to low-quality data. Existing demultiplexing methods can accurately assign donor identities in high-quality datasets, but they often fail on low-quality data. To address this, we developed CMDdemux, a method comprising three key steps: within-cell centered log-ratio (CLR) normalization of hashtag count data, K-medoids clustering, and classification of cells based on Mahalanobis distance. By integrating both hashing and mRNA data, CMDdemux achieves high accuracy in distinguishing singlets, doublets, and negatives. It also provides visualization tools to help users inspect potentially misclassified droplets. We benchmarked CMDdemux against existing methods using a range of high- and low-quality datasets. Results show that CMDdemux consistently outperforms other approaches, demonstrating robust performance on both high- and low-quality data where other methods fail. CMDdemux is particularly effective in handling diverse types of low-quality multiplexing data across different multiplexing technologies.

bioinformatics↗

bioRxiv2025

Three-dimensional patient-derived models of glioblastoma retain intra-tumoral heterogeneity

The intra- and inter-tumoral heterogeneity of glioblastoma represents a significant therapeutic challenge, as well as difficulty in generating reliable models for in vitro studies. Historical 2D adherent cell lines do not recapitulate this complexity, whereas both patient-derived neurospheres (PDN) and organoids (PDO) demonstrate intra-tumoral heterogeneity. Here, we quantify the tumor cell composition from matched models established from the same primary tumor using a series of multi-omic interrogations. We find that both patient-derived models recapitulate the genomic, epigenomic and tumor cell heterogeneity of the primary tissue. Furthermore, single-nuclei RNA sequencing revealed a subset of organoids containing small numbers of non-malignant cells from neuron and immune cell lineages. Harnessing the intra-tumoral heterogeneity of PDN models, we reveal the impact of temozolomide chemotherapy on individual cell states, altering composition of tumors over time in response to treatment. Our data confirms that both patient-derived models recapitulate patient intra-tumoral heterogeneity providing a platform for tumor cell state refined therapeutic studies. Key PointsO_LIGeneration of matched patient-derived neurosphere and organoid models from resected GBM tissue C_LIO_LINeurosphere models exhibit greater proliferative signatures C_LIO_LIBoth patient-derived models recapitulate genomic and epigenomic features of the primary tissue C_LIO_LISingle-nuclei RNA sequencing reveals both models recapitulate intra-tumoral heterogeneity of the primary tissue C_LIO_LINeurosphere models enable interrogation of therapeutic responses in the context of heterogeneity C_LI Importance of studyPatient derived models can be powerful tools when they faithfully recapitulate the tumor tissue from which they are derived. In glioblastoma, patient derived neurospheres (PDN) and organoids (PDO) have both been used in studies, however the differences between these models and the recapitulation of patient heterogeneity remain to be fully characterized. To address this, we performed multi-omic profiling of PDN and PDO models generated from the same tumor tissue. We find that that across a range of data modalities, both model systems exhibit a high level of resemblance to tissue, and critically, maintain heterogeneity and tumor cell composition. The importance of modeling heterogeneity was demonstrated in PDN models, where temozolomide treatment specifically alters the abundance of MES-like and AC-like tumor cells. Our findings demonstrate that neurosphere and organoid models effectively preserve cellular heterogeneity, genomic alterations, methylation signatures and transcriptomic features, both highly suitable to model glioblastomas complex cellular landscape.

cancer biology↗

bioRxiv2024

TIRE-seq: an Integrated Sample Extraction and Transcriptomics Workflow for High Throughput Perturbation Studies

RNA sequencing (RNA-seq) is widely used in biomedical research, advancing our understanding of gene expression across biological systems. Traditional methods require upstream RNA extraction from biological inputs, adding time and expense to workflows. We developed TIRE-seq (Turbocapture Integrated RNA Expression Sequencing) to address these challenges. TIRE-seq integrates mRNA purification directly into library preparation, eliminating the need for a separate extraction step. This streamlined approach reduces turnaround time, minimizes sample loss, and improves data quality. A comparative study with the widely used Prime-seq protocol demonstrates TIRE-seqs superior sequencing efficiency with crude cell lysates as inputs. TIRE-seqs utility was demonstrated across three biological applications. It captured transcriptional changes in stimulated human T cells, revealing activation-associated gene expression profiles. It also identified key genes driving murine dendritic cell differentiation, providing insights into lineage commitment. Lastly, TIRE-seq analyzed the dose-response and time-course effects of temozolomide on patient-derived neurospheres, identifying differentially expressed genes and enriched pathways linked to the drugs mechanism of action. With its simplified workflow and high sequencing efficiency, TIRE-seq offers a cost-effective solution for large-scale gene expression studies across diverse biological systems.

genomics↗

bioRxiv2023

scPipe: An extended preprocessing pipeline for comprehensive single-cell ATAC-Seq data integration in R/Bioconductor

scPipe is a flexible R/Bioconductor package originally developed to analyse platform-independent single-cell RNA-Seq data. To expand its preprocessing capability to accommodate new single-cell technologies, we further developed scPipe to handle single-cell ATAC-Seq and multi-modal (RNA-Seq and ATAC-Seq) data. After executing multiple data cleaning steps to remove duplicated reads, low abundance features and cells of poor quality, a SingleCellExperiment object is created that contains a sparse count matrix with features of interest in the rows and cells in the columns. Quality control information (e.g. counts per cell, features per cell, total number of fragments, fraction of fragments per peak) and any relevant feature annotations are stored as metadata. We demonstrate that scPipe can efficiently identify "true" cells and provides flexibility for the user to fine-tune the quality control thresholds using various feature and cell-based metrics collected during data preprocessing. Researchers can then take advantage of various downstream single-cell tools available in Bioconductor for further analysis of scATAC-Seq data such as dimensionality reduction, clustering, motif enrichment, differential accessibility and cis-regulatory network analysis. The scPipe package enables a complete beginning-to-end pipeline for single-cell ATAC-Seq and RNA-Seq data analysis in R.

bioinformatics↗

bioRxiv2023

A Risk-reward Examination of Sample Multiplexing Reagents for Single Cell RNA-Seq

Single-cell RNA sequencing (scRNA-Seq) has emerged as a powerful tool for understanding cellular heterogeneity and function. However the choice of sample multiplexing reagents can impact data quality and experimental outcomes. In this study, we compared various multiplexing reagents, including MULTI-Seq, Hashtag antibody, and CellPlex, across diverse sample types such as human peripheral blood mononuclear cells (PBMCs), mouse embryonic brain and patient-derived xenografts (PDXs). We found that all multiplexing reagents worked well in cell types robust to ex vivo manipulation but suffered from signal-to-noise issues in more delicate sample types. We compared multiple demultiplexing algorithms which differed in performance depending on data quality. We find that minor improvements to laboratory workflows such as titration and rapid processing are critical to optimal performance. We also compared the performance of fixed scRNA-Seq kits and highlight the advantages of the Parse Biosciences kit for fragile samples. Highly multiplexed scRNA-Seq experiments require more sequencing resources, therefore we evaluated CRISPR-based destruction of non-informative genes to enhance sequencing value. Our comprehensive analysis provides insights into the selection of appropriate sample multiplexing reagents and protocols for scRNASeq experiments, facilitating more accurate and cost-effective studies.

genomics↗

bioRxiv2022

Multimodal single cell analysis of the paediatric lower airway reveals novel immune cell phenotypes in early life health and disease

Inflammation is a key driver of cystic fibrosis (CF) lung disease, not addressed by current standard care. Improved understanding of the mechanisms leading to aberrant inflammation may assist the development of effective anti-inflammatory therapy. Single-cell RNA sequencing (scRNA-seq) allows profiling of cell composition and function at previously unprecedented resolution. Herein, we seek to use multimodal single-cell analysis to comprehensively define immune cell phenotypes, proportions and functional characteristics in preschool children with CF. We analyzed 42,658 cells from bronchoalveolar lavage of 11 preschool children with CF and a healthy control using scRNA-seq and parallel assessment of 154 cell surface proteins. Validation of cell types identified by scRNA-seq was achieved by assessment of samples by spectral flow cytometry. Analysis of transcriptome expression and cell surface protein expression, combined with functional pathway analysis, revealed 41 immune and epithelial cell populations in BAL. Spectral flow cytometry analysis of over 256,000 cells from a subset of the same patients revealed high correlation in major cell type proportions across the two technologies. Macrophages consisted of 13 functionally distinct sub populations, including previously undescribed populations enriched for markers of vesicle production and regulatory/repair functions. Other novel cell populations included CD4 T cells expressing inflammatory IFN/{beta} and NF{kappa}B signalling genes. Our work provides a comprehensive cellular analysis of the pediatric lower airway in preschool children with CF, reveals novel cell types and provides a reference for investigation of inflammation in early life CF.

systems biology↗