Biology subjects

Tan, A.

Publications and source records attributed to Tan, A..

At least 19 recordsLinked to original sources

bioRxiv2026

Comparative analysis of adhesin-like proteins from Methanobacteriales species

Methanogenic archaea are key syntrophic partners in animal and human guts. However, the molecular basis of physical and functional interactions of methanogens with other cells remains poorly understood. Adhesin-like proteins (ALPs) are abundant, unusually large, and repeat-rich surface proteins and are potential prime candidates for mediating cell-cell interactions, however, due to poor characterization, ALPs often remain neglected in annotations of genomes and large-scale datasets. Here, we combined structure-guided analysis with comparative genomics to map ALP domain architectures across Methanobacteriales species, with emphasis on intestinal and rumen lineages. We curated predicted ALP structures and manually delineated domain boundaries to build a reference set encompassing ABD, RBH, membrane-anchoring domains (MAD), and miscellaneous accessory domains. Leveraging on a novel approach for ALP domain annotations, we generate ALP annotations and domain architectures for 17 Methanobacteriales species without reliance on primary-sequence homology. Applying this approach to strains of Methanobrevibacter smithii and Methanobrevibacter intestini, we assembled species-level "adhesiomes" and assessed conservation patterns of ALPs. Network analysis indicated that ALPs are more commonly shared among strains within a species than between species, suggesting species-specific adhesin repertoires. While total ALP counts differed significantly between M. smithii and M. intestini (Wilcoxon rank-sum p = 0.00295), median ALP lengths did not (p = 0.606). Our results provide a comprehensive overview of the ALP domain architectures in Methanobacteriales species and reveals species-specific adhesiomes that likely contribute to niche adaptation and syntrophic partnerships in gut ecosystems.

microbiology↗

bioRxiv2026

Systematic benchmarking of multi-modal approaches for tumor-naive ctDNA detection and quantification

Longitudinal monitoring of circulating tumor DNA (ctDNA) has emerged as a promising framework for characterizing treatment response dynamics in cancer. Scalable tumor-naive approaches for quantifying ctDNA often involve whole-genome sequencing (WGS) or DNA methylation profiling, but their comparative performance and capacity for complementary integration remain poorly understood. Here we systematically benchmarked tumor-naive WGS- and methylation-based ctDNA quantification methods using plasma from 150 patients with colorectal, lung and breast cancer. Using paired high-depth WGS and EM-seq data, we generated 40,000 in silico samples and evaluated detection accuracy, limits of detection (LoD) and quantification (LoQ) across cancer types and sequencing depths (0.1x-30x). We further assessed single- and multimodal method combinations, identifying conditions under which integrated approaches enhance analytical performance for detection and quantification relative to single modalities. This benchmark delineates key performance trade-offs and provides a practical framework to support method development and guide future research applications in ctDNA-based biomarker studies.

bioinformatics↗

bioRxiv2026

Sequence-independent protein domain detection and classification with PRISM

The explosion of predicted protein structures has revealed countless novel domain families. However, gold-standard segmentation tools like Chainsaw and Merizo are trained on rapidly obsoleting CATH databases, lack automatic domain classification, and cannot be easily fine-tuned without deep learning expertise. We introduce PRISM, a unified framework enabling sequence-independent, one-shot fine-tuning for simultaneous domain segmentation and classification, bypassing traditional constraints to accurately resolve complex, novel protein architectures.

bioinformatics↗

bioRxiv2026

Phylogenetic Analysis and Structural Evaluation of Staphylococcus aureus Serine-Aspartate Repeat-Containing Protein D with a Focus on Periprosthetic Joint Infection

Serine-aspartate repeat-containing protein D (SdrD) is a Staphylococcus aureus cell wall-anchored, calcium-binding adhesin member of the MSCRAMM Sdr subfamily that may contribute to bacterial adhesion and virulence. S. aureus is the most common cause of periprosthetic joint infection (PJI). Population-level distribution and sequence diversity of SdrD among clinical PJI isolates have not been systematically characterized, and the SdrD binding mechanism is still not well understood. To address these gaps, sdrD alleles were queried across 156 newly sequenced PJI isolates and compared to publicly available S. aureus genomes, and nucleotide- and protein-level phylogenies of the sdrCDE locus constructed. The SdrD crystal structure from S. aureus JH1 was determined, with solution small-angle X-ray scattering (SAXS) and molecular dynamics (MD) simulations, and assessment of conformational changes with calcium depletion. Three dominant sdrD subtypes were defined, associating with USA300, JH1, and TCH60; the JH1 sdrD subtype was predominant among PJI isolates. Structural studies showed that the conformation of individual domains and interdomain organization of the multidomain SdrD have limited flexibility in solution, and that the calcium-binding B domain retains its core fold under conditions of calcium depletion. Together, the findings presented support functional diversification among Sdr family members in mediating host attachment and inform a re-evaluation of the ligand-binding mechanism previously proposed for SdrD. AUTHOR SUMMARYStaphylococcus aureus is the leading cause of infections that develop around joint implants (periprosthetic joint infection, PJI). This bacterium has a large arsenal of surface proteins that allow it to stick to human tissues and implanted devices. This work focused on one such protein, SdrD, which has been linked to implant-associated infections but the structure and diversity of which among patients with PJI had not been well characterized. The genetic sequences of SdrD were analyzed across thousands of bacterial genomes, including those from patients with PJI. Distinct genetic variants of the protein were found, one of which was particularly common with PJI. The three-dimensional structure of SdrD was determined at atomic resolution and solution small-angle X-ray scattering (SAXS) and molecular dynamics used to study how it moves and responds to changes in its environment. Contrary to what was previously described, SdrD was shown to be relatively rigid. These findings change how SdrDs mechanism of action should be considered, potentially informing design strategies to block bacterial attachment before infection takes hold.

biophysics↗

bioRxiv2026

Reward Circuit Adaptations After Chronic Antipsychotic Treatment Confer Addiction Vulnerability

ObjectiveAntipsychotics are widely used chronically to treat several psychiatric conditions and off-label indications, but their clinical utility is often accompanied by motor and/or metabolic side effects. Antipsychotics can also alter the brains reward circuitry, and we recently showed that upon abrupt discontinuation, chronic treatment with the first-generation antipsychotic haloperidol increases cocaine relapse in a rodent model of i.v. cocaine self-administration. Here, we examined whether chronic treatment with second-generation antipsychotics was also associated with increased addiction vulnerability in rodents and humans, and sought to further characterize the cellular substrate linking antipsychotic exposure to substance use disorder (SUD). MethodsAdult rats received subcutaneous infusion of risperidone, olanzapine, or vehicle for 14 days. Rats were then trained to self-administer i.v. cocaine or oral sucrose pellets for 10 days and active lever presses were paired with compound cue exposure. After a period of extinction training, reward seeking was reinstated by restoration of cues to active lever pressing for 20 minutes. To enable quantification of dendritic spine morphology and manipulation of MSNs during relapse, viral approaches were used to label D1- or D2-MSNs in the nucleus accumbens core (NAcore), a region that encodes reward-related responses and regulates symptoms associated with addiction and other psychiatric disorders, including schizophrenia. To assess the clinical relationship between chronic antipsychotic use and SUD, we conducted a systematic review and meta-analysis of studies identified through PubMed, CENTRAL, Scopus, and Embase that examined the effects of antipsychotic treatment on addictive substance use or craving in patients with SUD. ResultsIn rats, chronic risperidone or olanzapine treatment delayed extinction learning and increased cue-induced cocaine seeking, but did not alter seeking of the natural reward sucrose or its extinction. Cued relapse increased spine head diameter in NAcore D2-MSNs, but not D1-MSNs after antipsychotic treatment relative to vehicle-treated controls. Chemogenetic inhibition of D2-MSNs reversed addiction vulnerability conferred by antipsychotic treatment. In patients whose SUDs were treated with antipsychotics, duration of treatment (>4 months) strongly predicted addictive substance use or craving independent of antipsychotic type (first- vs. second-generation) or misused substance (psychostimulant, opioid, alcohol, or cannabis). ConclusionsChronic antipsychotic treatment, regardless of drug class, was associated with increased vulnerability to addictive substance use and relapse-like behavior in a rat model, and was linked to potentiation of D2-MSNs in the ventral striatum. In humans, systematic review and meta-analytic findings supported this pattern in patients maintained on antipsychotic treatment.

neuroscience↗

bioRxiv2026

New insights into the evolution of spider silk proteins illuminated by long-read transcriptomes

Spider silk has long fascinated scientists because of its exceptional mechanical properties, yet the evolutionary origins and diversification of spider silk proteins (spidroins) remain incompletely understood. By analyzing long-read spidroin transcripts from spiders representing key evolutionary lineages, we identify two ancestral proteins present in basal spiders: an alanine-serine-rich (AS-type) protein and a glycine-serine-rich (GS-type) protein. These ancestral proteins likely served as primary evolutionary templates for the diversification of modern spider silks. We show that the AS-type ancestral spidroin remained relatively conserved and ultimately gave rise to the specialized tubuliform spidroin (TuSp) used in eggcase silk. In contrast, the GS-type protein underwent extensive functional radiation, evolving into differentiated minor ampullate spidroins (MiSp) through the acquisition of distinct terminal domains and {beta}-sheet-associated structural innovations. Our results further suggest that major ampullate spidroins (MaSp) may have originated from MiSp, with the GS motif representing an evolutionarily favorable genetic substrate for the emergence of high-performance silk proteins. Finally, we propose a revised evolutionary trajectory for flagelliform spidroins (Flag), suggesting that they were co-opted as components of ampullate silk during functional degeneration in modern non-web-weaving RTA clade spiders. Together, these findings provide a high-resolution framework for understanding the genetic innovations that drove the diversification of spider silk.

evolutionary biology↗

bioRxiv2025

RiboZAP: Custom Probe Design for rRNA Depletion in complex Metatranscriptomes

Metatranscriptomic (MetaT) sequencing provides critical insights into the gene expression and functional activity of microbial communities. However, its utility is limited by the overwhelming abundance of ribosomal RNA (rRNA), which typically represents [≥]90% of total RNA [1-2]. A major obstacle to efficient MetaT analysis is the removal of highly abundant rRNA transcripts present in complex microbial communities, which may contain thousands of species. Although commercial rRNA depletion kits can effectively reduce rRNA content, they are typically optimized for specific host microbiomes and often underperform in others. For example, probes designed for the human gut microbiome frequently show reduced efficiency when applied to non-human samples such as mouse cecal donor samples - a common model in microbiome research. Regardless of the depletion strategy used, designing rRNA removal probes solely based on a microbiomes taxonomic composition often requires an extensive number of probes, making the approach expensive, difficult to manufacture, and sometimes technically impractical [3]. Here, we present RiboZAP, a species-agnostic computational pipeline for designing custom RNase H depletion probes directly from MetaT sequencing data and without prior knowledge of sample composition. Our results show that the probes generated with RiboZAP are efficient for removal of rRNA content, increasing messenger RNA (mRNA), and improving transcriptome coverage. This provides a cost-effective approach to maximize the value of MetaT sequencing.

bioinformatics↗

bioRxiv2025

Gene-Family Encoding Boosts Domain-Adapted Single-Cell Language Models

Transformer-based single-cell foundation models often rely on ranked-gene (RG) sequences where genes, ranked by expression, are often not functionally related, weakening next-token learning and the structure of learned embeddings. Here, we introduce gene-family (GF) encoding, where expressed genes are grouped into functionally defined families, and ranking performed within each family. Using 100,000 gastric-cancer (GC) cells, we domain-adapted 8-billion-parameter Llama and Qwen backbones with either RG or GF sentences and benchmarked zero-shot performance on embedding-based and generative tasks. GF models outperformed RG models across both tasks and backbones. We scaled GF-Llama to 1.3 million GC cells to obtain GF-Llama-GC and applied it to two applications: resolving fine-grained cellular heterogeneity and discovering cell populations associated with disease progression. GF-Llama-GC revealed immune-cell subclusters not resolved by standard expression-based analyses. Applying in-silico cell removal/transplantation on a chemotherapy responder/non-responder scRNA-seq dataset, GF-Llama-GC highlighted not only epithelial cells but also neutrophils as key cells associated with chemotherapy response.

bioinformatics↗

bioRxiv2025

CHIEF: An Attention-based Ensemble Learning Framework for Functional Protein Design

Protein de novo design is a longstanding challenge in biological field due to the multifaceted physicochemical properties and complexities of proteins. Several deep learning models have emerged for protein de novo design. However, each model exhibits distinct strengths and limitations because of the differences in training strategies or neural network architectures. Therefore, improving the effectiveness of protein design, particularly for functional protein, requires further improvement. To tackle this challenge, we developed CHIEF (Chimera Ensemble Inverse Folding), an attention-based ensemble framework that integrates five pretrained base models (ProteinMPNN-vanilla, ProteinMPNN-soluble, ESM-IF, Frame2seq and PiFold) to leverage their complementary strengths and mitigate limitations. Compared with the base models, CHIEF significantly improved the sequence recovery rate by 16.6-28.0%, while reducing the prediction perplexity by 22.7-34.6%. CHIEF also exhibited superior protein-designing capacity in variable sequence- and structure-based metrics and in large and complex proteins. CHIEF dynamically captured the semantic information from base models in a context-dependent manner and was minimally affected by the ablation of each base model. More importantly, CHIEF demonstrates real-world applicability to design functional malate dehydrogenase (MDH), achieving a 100% success rate. In summary, our study develops an ensemble deep learning model that improves the efficacy of protein sequence design, and will be a valuable platform for protein engineering and drug development.

bioengineering↗

bioRxiv2025

Optimizing mouse metatranscriptome profiling by selective removal of redundant nucleic acid sequences

Metatranscriptome (MetaT) sequencing is a critical tool for profiling the dynamic metabolic functions of microbiomes. In addition to taxonomic information, MetaT also provides real-time gene expression data of both host and microbial populations, thus permitting authentic quantification of the functional (enzymatic) output of the microbiome and its host. The main challenge to effective and accurate MetaT analysis is the removal of highly abundant rRNA transcripts from these complex mixtures of microbes, which can number in the thousands of individual species. Regardless of methodology for rRNA depletion, the design of rRNA removal probes based solely upon taxonomic content of the microbiome typically requires very large numbers of individual probes, making this approach complex to commercially manufacture, costly, and frequently technically infeasible. In previous work [1], we designed a set of depletion probes for human stool samples using a design strategy based solely on sequence abundance, completely agnostic of the microbiomal species present. Here, we show that the human-based probes are less effective when used with mouse cecal samples. However, adapting additional rRNA depletion probes specifically to cecal content provides both greater efficiency and consistency for MetaT analysis of mouse samples. ImportanceSequencing total RNA from microbiome samples is seriously impaired by the overwhelming proportion of rRNA to mRNA content. As much as 99% of sequencing reads can be assigned to the rRNA content, thus removal of these abundant transcripts is critical to MetaT analysis. The use of Ribo Zero Plus rRNA depletion probes designed for human gut microbiomes proved to be less effective and more inconsistent across mouse cecal donor samples, a common experimental system for microbiome studies. In the present work, we have extended and refined a taxonomically-neutral probe design method for mouse cecal content. The additional probes were carefully chosen to limit the number needed for effective depletion to reduce both the cost and risk of introducing bias to MetaT analysis. Our results demonstrate this method as efficient and consistent for rRNA removal in mouse cecal samples, thus providing a significant increase in the number of mRNA-rich sequencing reads for MetaT analysis.

genomics↗

bioRxiv2024

Evolving a High-Performance Terminal Deoxynucleotidyl Transferase for Enzymatic DNA Synthesis

Enzymatic DNA synthesis, using stepwise nucleotide addition catalyzed by template-independent polymerases, promises higher efficiency, quality, and sustainability than todays industry standard phosphoramidite-based processes. We report on the directed evolution of a terminal deoxynucleotidyl transferase that uses 3-phosphate blocked dNTPs to control the polymerization reaction and demonstrates high activity for these modified substrates and improved template promiscuity and thermostability.

bioengineering↗

bioRxiv2023

Multiple performance peaks for scale-biting in an adaptive radiation of pupfishes

The physical interactions between organisms and their environment ultimately shape their rate of speciation and adaptive radiation, but the contributions of biomechanics to evolutionary divergence are frequently overlooked. Here we investigated an adaptive radiation of Cyprinodon pupfishes to measure the relationship between feeding kinematics and performance during adaptation to a novel trophic niche, lepidophagy, in which a predator removes only the scales, mucus, and sometimes tissue from their prey using scraping and biting attacks. We used high-speed video to film scale-biting strikes on gelatin cubes by scale-eater, molluscivore, generalist, and hybrid pupfishes and subsequently measured the dimensions of each bite. We then trained the SLEAP machine-learning animal tracking model to measure kinematic landmarks and automatically scored over 100,000 frames from 227 recorded strikes. Scale-eaters exhibited increased peak gape and greater bite length; however, substantial within-individual kinematic variation resulted in poor discrimination of strikes by species or strike type. Nonetheless, a complex performance landscape with two distinct peaks best predicted gel-biting performance, corresponding to a significant nonlinear interaction between peak gape and peak jaw protrusion in which scale-eaters and their hybrids occupied a second performance peak requiring larger peak gape and greater jaw protrusion. A bite performance valley separating scale-eaters from other species may have contributed to their rapid evolution and is consistent with multiple estimates of a multi-peak fitness landscape in the wild. We thus present an efficient deep-learning automated pipeline for kinematic analyses of feeding strikes and a new biomechanical model for understanding the performance and rapid evolution of a rare trophic niche.

evolutionary biology↗

bioRxiv2023

Comparison of tet(X4)-containing contigs from metagenomic sequencing data with plasmid sequences of isolates from a cohort of healthy subjects.

The recently discovered tigecycline-inactivating enzyme Tet(X4) can confer high-level tigecycline resistance on its hosts, which makes it a public health concern. This study focused on detection, analysis, and characterization of Tet(X4)-positive Enterobacteriaceae from the gut microbiota of a healthy cohort of individuals in Singapore using cultivation-dependent and cultivation-independent approaches. Twelve Tet(X4)-positive Enterobacteriaceae strains that were previously obtained from the cohort were fully genome-sequenced and comparatively analysed. A metagenomic sequencing (MS) dataset of the same samples was mined for contigs that harboured the tet(X4) resistance gene. The sequences of tet(X4)-containing contigs and plasmids sequences were compared. The presence of the resistance genes floR and catD (also annotated as estT) was detected in the same cassette in 10 and 12 out of the 12 tet(X4)-carrying plasmids, respectively. MS detected tet(X4)-containing contigs in two out of 109 subjects, while cultivation-dependent analysis previously reported a prevalence of 10.1%. Contig sequences are relatively short ([~]14-33 kb) but show high similarity to the respective plasmid sequences of the isolates. The frequent co-occurrence of florR and catD with tet(X4) corroborates the hypothesis that the transmission of tet(X4) may have originated from the veterinary sector. Our findings show that MS can complement efforts in the surveillance of antibiotic resistance genes for clinical samples, while it has a lower sensitivity than a cultivation-based method when the target organism have a low abundance. Further optimisation is required if MS is to be utilised in antibiotic resistance surveillance.

microbiology↗

bioRxiv2023

The Host-specific Microbiota is Required for Diet-Specific Metabolic Homeostasis

In complex mammals, the importance and host-specificity of microbial communities have been demonstrated through their positive effects on host immune fitness or performance. However, whether host metabolic physiology homeostasis depends on a specific bacterial community exclusive to the host remains unclear. Here, we show that the coevolved host-specific microbiota is required to maintain diet-specific flexible and sufficient metabolic homeostasis through a high colonization rate, modulating gut metabolites, and related targets. Using germ-free (GF) mice, we tested whether the fitness benefiting the host metabolic phenotype of microbiota was host-specific. We demonstrated that GF mice associated with exogenous microbiota (human microbiota (HM)), which exhibited different and reduced gut microbial species diversity, significantly elevated metabolic rate, and exhibited metabolic insufficiency, all characteristics of GF mice. Strikingly, the absence of the host-specific microbiome attenuated high-fat diet-specific metabolism features. Different diets caused different metabolic changes in only host-specific microbiota-associated mice, not the host-microbiota mismatched mice. While RNA sequencing revealed subtle changes in the expression of genes in the liver, GF mice and HM mice showed considerably altered expression of genes associated with metabolic physiology compared to GF mice associated with host-specific microbiota. The effect of diet outweighed microbiota in the liver transcriptome. These changes occurred in the setting of decreased luminal short-chain fatty acids (SCFAs) and the secondary bile acid (BAs) pool and downstream gut signaling targets in HM and GF mice, which affects whole-body metabolism. These data indicate that a foreign microbial community provides little metabolic benefit to the host when compared to a host-specific microbiome, due to the colonization selection pressure and microbiota-derived metabolites dysfunction. Overall, microbiome fitness effects on the host metabolic phenotype were host-specific. Understanding the impact of the host-specificity of the microbiome on metabolic homeostasis may provide important insights for building a better probiotic. HighlightsO_LIMicrobiome fitness effects on the host metabolic phenotype were host-specific in mammals. C_LIO_LIHuman microbiota-associated mice exhibited lower host metabolic fitness or performance, and similar functional costs in GF mice. C_LIO_LIDifferent diets cause different metabolic changes only in host-specific microbiota-associated mice, not the host-microbiota mismatched mice. C_LIO_LIThe defective gut microbiota in host-specific microbiota, microbial metabolites and related targets likely drive the metabolic homeostasis. C_LI Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=132 SRC="FIGDIR/small/565654v1_ufig1.gif" ALT="Figure 1"> View larger version (41K): org.highwire.dtl.DTLVardef@16a864org.highwire.dtl.DTLVardef@1894996org.highwire.dtl.DTLVardef@3e5429org.highwire.dtl.DTLVardef@77bfb_HPS_FORMAT_FIGEXP M_FIG C_FIG

microbiology↗

bioRxiv2023

Broad variation in response of individual introns to splicing inhibitors in a humanized yeast strain

Intron branch point (BP) recognition by the U2 snRNP is a critical step of splicing, vulnerable to recurrent cancer mutations and bacterial natural product inhibitors. The BP binds a conserved pocket in the SF3B1 (human) or Hsh155 (yeast) U2 snRNP protein. Amino acids that line this pocket affect binding of splicing inhibitors like Pladienolide-B (Plad-B), such that organisms differ in their sensitivity. To study the mechanism of splicing inhibitor action in a simplified system, we modified the naturally Plad-B resistant yeast Saccharomyces cerevisiae by changing 14 amino acids in the Hsh155 BP pocket to those from human. This humanized yeast grows normally, and splicing is largely unaffected by the mutation. Splicing is inhibited within minutes after addition of Plad-B, and different introns appear inhibited to different extents. Intron-specific inhibition differences are also observed during co-transcriptional splicing in Plad-B using single-molecule intron tracking (SMIT) to minimize gene-specific transcription and decay rates that cloud estimates of inhibition by standard RNA-seq. Comparison of Plad-B intron sensitivities to those of the structurally distinct inhibitor Thailanstatin-A reveals intron-specific differences in sensitivity to different compounds. This work exposes a complex relationship between binding of different members of this class of inhibitors to the spliceosome and intron-specific rates of BP recognition and catalysis. Introns with variant BP sequences seem particularly sensitive, echoing observations from mammalian cells, where monitoring individual introns is complicated by multi-intron gene architecture and alternative splicing. The compact yeast system may hasten characterization of splicing inhibitors, accelerating improvements in selectivity and therapeutic efficacy.

molecular biology↗

bioRxiv2023

A transcriptome-wide meta-analysis reveals lack of cancer-cell intrinsic determinants of response to immune checkpoint blockade

Immune-checkpoint therapy (ICB) has conferred significant and durable clinical benefit to some cancer patients. However, most patients do not respond to ICB, and reliable biomarkers of ICB response are needed to improve patient stratification. Here, we performed a transcriptome-wide meta-analysis across 1,486 tumors from ICB-treated patients and tumors with expected ICB outcomes based on microsatellite status. Using a robust transcriptome deconvolution approach, we inferred cancer and stroma-specific gene expression differences and identified cell-type specific features of ICB response across cancer types. Consistent with current knowledge, stromal expression of CXCL9, CXCL13, and IFNG were the top determinants of favorable ICB response. In addition, we identified a group of potential immune-suppressive genes, including FCER1A, associated with poor response to ICB. Strikingly, PD-L1 expression in stromal cells, but not cancer cells, is correlated with ICB response across cancer types. Furthermore, the unbiased transcriptome-wide analysis failed to identify cancer-cell intrinsic features of ICB response conserved across tumor types. Overall, our results challenge the prevailing dogma that cancer cells present tissue-agnostic molecular markers that modulate immune activity and ICB treatment response. These results have implications for the development of improved ICB treatments and diagnostics.

cancer biology↗

bioRxiv2023

nELISA: A high-throughput, high-plex platform enables quantitative profiling of the secretome

We present the nELISA, a high-throughput, high-fidelity, and high-plex protein profiling platform. DNA oligonucleotides are used to pre-assemble antibody pairs on spectrally encoded microparticles and perform displacement-mediated detection. Spatial separation between non-cognate antibodies prevents the rise of reagent-driven cross-reactivity, while read-out is performed cost-efficiently and at high-throughput using flow cytometry. nELISA can measure both protein concentration and their post-translational modifications. We assembled an inflammatory panel of 191 targets that were multiplexed without cross-reactivity nor impact on performance vs 1-plex signals, with sensitivities as low as 0.1 pg/mL and measurements spanning 7 orders of magnitude. We then performed a large-scale inflammatory-secretome perturbation screen of peripheral blood mononuclear cells (PBMCs), with cytokines as both perturbagens and read-outs, measuring 7,392 samples and generating [~]1.4M protein data points in under a week; a significant advance in throughput compared to other highly multiplexed immunoassays. We uncovered 447 significant cytokine responses, including multiple putatively novel ones, that were conserved across donors and stimulation conditions. We validate nELISA for phenotypic screening, where its capacity to faithfully report hundreds of proteins make it a powerful tool across multiple stages of drug discovery.

bioengineering↗

bioRxiv2023

Using photographs and deep neural networks to understand flowering phenology and diversity in mountain meadows

Mountain meadows are an essential part of the alpine-subalpine ecosystem; they provide ecosystem services like pollination and are home to diverse plant communities. Changes in climate affect meadow ecology on multiple levels, for example by altering growing season dynamics. Tracking the effects of climate change on meadow diversity through the impacts on individual species and overall growing season dynamics is critical to conservation efforts. Here, we explore how to combine crowd sourced camera images with machine learning to quantify flowering species richness across a range of elevations in alpine meadows located in Mt Rainier National Park, Washington, USA. We employed three machine learning techniques (Mask R-CNN, RetinaNet and YOLOv5) to detect wildflower species in images taken during two flowering seasons. We demonstrate that deep learning techniques can detect multiple species, providing information on flowering richness in photographed meadows. The results indicate higher richness just above the tree line for most of the species, which is comparable with patterns found using field studies. We found that the two-stage detector Mask R-CNN was more accurate than single-stage detectors like RetinaNet and YOLO, with the Mask R-CNN network performing best overall with mean average precision (mAP) of 0.67 followed by RetinaNet (0.5) and YOLO (0.4). We found that across the methods using anchor box variations in multiples of 16 led to enhanced accuracy. We also show that detection is possible even when pictures are interspersed with complex backgrounds and are not in focus. We found differential detection rates depending on species abundance, with additional challenges related to similarity in flower characteristics, labeling errors, and occlusion issues. Despite these potential biases and limitations in capturing flowering abundance and location-specific quantification, accuracy was notable considering the complexity of flower types and picture angles in this data set. We therefore expect that this approach can be used to address many ecological questions that benefit from automated flower detection, including studies of flowering phenology and floral resources, and that this approach can therefore complement a wide range of ecological approaches (e.g., field observations, experiments, community science, etc.). In all, our study suggests that ecological metrics like floral richness can be efficiently monitored by combining machine learning with easily accessible publicly curated datasets (e.g., Flickr, iNaturalist).

ecology↗