Biology subjects

Iyer, A.

Publications and source records attributed to Iyer, A..

At least 19 recordsLinked to original sources

bioRxiv2026

Benchmarking Large Language Models for Transcriptomic Data Analysis and Visualization: Multi-Model Workflow Evaluation

This study presents a comprehensive benchmarking evaluation of large language models(LLMs) and agentic coding systems for automated analysis and visualization workflows for transcriptomic data, in particular human responses to COVID-19 infection. Multiple models, including ChatGPT 5.5, Claude Sonnet 4.6, Gemini 3.1, and Llama4-Scout, were assessed across a series of scientific prompts designed to test their ability to reproduce workflow components such as dimensionality reduction, clustering, outlier detection, reproducibility, and code generation. The benchmark protocol involved varying prompt specificity and examined both quantitative metrics such as clustering scores and outlier counts as well as qualitative aspects such as code quality and reproducibility. Results indicate that prompt specificity improves code structure and methodological correctness, but quantitative performance depends heavily on preprocessing choices and execution environment. While some models produced strong code and reproducible outputs, alignment with scientific targets like label-based clustering was inconsistent. The findings highlight the promise and limitations of current LLMs for automating scientific data analysis and visualization and suggest directions for future benchmarking and workflow standardization.

bioinformatics↗

bioRxiv2026

Ms4a4a loss preferentially remodels microglia with limited, compartment-specific effects across central and peripheral immune compartments in 5xFAD mice

Background: Genome-wide association studies have identified the MS4A locus as a major modifier of Alzheimers disease (AD) risk and resilience. MS4A4A is enriched in microglia, and we previously demonstrated that loss of Ms4a4a reduces amyloid plaque pathology and attenuates amyloid-associated tau spreading. However, MS4A4A is also expressed in peripheral myeloid populations, raising the possibility that its effects extend beyond brain-resident microglia. Here, we sought to define how constitutive Ms4a4a loss alters immune populations across central nervous system (CNS), CNS-border, and peripheral compartments in the context of amyloid pathology. Methods: We performed multicompartment immune profiling by flow cytometry in 5xFAD Ms4a4a wild-type and 5xFAD Ms4a4a-deficient mice. Myeloid and lymphoid populations were characterized in the brain, dural meninges, skull bone marrow, tibial bone marrow, and spleen. Results: Ms4a4a loss produced its strongest immune effects in brain microglia. 5xFAD Ms4a4a-deficient mice exhibited increased microglial abundance, increased frequency and expression of CD86, increased MHC class II expression intensity, and enrichment of CXCR4-high microglia, whereas P2RY12 and TREM2 were unchanged. Brain macrophages showed increased MHC class II expression, whereas conventional dendritic cells (cDCs) in the brain showed reduced MHC class II expression, with limited changes in other infiltrating immune populations. Immune composition in the dura and skull bone marrow was largely preserved, although MHC class II expression was increased in skull bone marrow cDCs. Peripheral immune composition in tibial bone marrow and spleen was similarly preserved, with selective increases in M2-like macrophages in tibial bone marrow and splenic CD8 T-cell frequency. Conclusions: These findings identify microglia as the immune population most strongly remodeled by Ms4a4a loss in the setting of amyloid pathology. Constitutive Ms4a4a deficiency does not produce broad systemic immune remodeling but instead preferentially alters microglial state, accompanied by more limited, cell type- and tissue-specific effects across other central and peripheral immune compartments. These findings establish MS4A4A as a context-dependent regulator of immune responses to AD pathology with particularly prominent effects on microglia.

neuroscience↗

bioRxiv2026

Ultrasound Activated Nanobubbles Induce Durable Systemic Antitumor Immunity

Clinical outcomes in aggressive breast cancer vary widely, in part because the tumor microenvironment is structured to exclude immune infiltration. Low antigen load, dysfunctional antigen-presenting cells, T cell exclusion and exhaustion, and a stiff extracellular matrix that physically restricts immune cell trafficking work together to form a suppressive barrier that current immunotherapies struggle to overcome. We addressed this barrier using ultrasound (US)-activated nanobubbles (NBs), a drug-free intervention based on perfluoropropane-filled nanoparticles. The size and deformable phospholipid shell enable NBs to achieve deep tumor penetration and a uniform distribution throughout the entire tumor. Upon ultrasound activation, NBs generate localized mechanical forces that restore extracellular matrix elasticity, disrupt tumor transport barriers, and drive HMGB1 release, re-engaging endogenous antitumor immunity without pharmacological agents. In a syngeneic triple-negative breast cancer model, US-NB treatment depleted immunosuppressive myeloid cells 3-fold within 3 hours, followed by a greater than 5-fold increase in the ratio of antigen-experienced to suppressive T cells at 48 hours. US-NB drives rapid infiltration of CD4+ and CD8+ T cells within 48 hours. US-NB treatment achieved an 85% cure rate in the D2A1 model; cured animals maintained durable systemic immune memory, rejecting both local and systemic tumor rechallenge. Consistent therapeutic benefit was observed in a luminal B-like mammary tumor model (E0771), supporting activity across breast cancer subtypes. These results establish US-NB mechanical immunomodulation as a drug-free therapeutic strategy capable of generating robust and durable antitumor immunity, acting through biophysical tissue properties rather than tumor-specific molecular targets. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=90 SRC="FIGDIR/small/714247v1_ufig1.gif" ALT="Figure 1"> View larger version (56K): org.highwire.dtl.DTLVardef@8f274eorg.highwire.dtl.DTLVardef@5da575org.highwire.dtl.DTLVardef@19defborg.highwire.dtl.DTLVardef@7e630e_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioengineering↗

bioRxiv2026

BRIDGE: Biological Antimicrobial Resistance Inference viaDomain-Knowledge Graph Embeddings

Antimicrobial resistance (AMR) is a growing global health crisis, responsible for an estimated 1.27 million deaths in 2019 alone. Traditional approaches to identifying antibiotic resistance genes (ARGs) are often labour-intensive and limited in their ability to detect novel resistance mechanisms. In this study, we propose BRIDGE, a knowledge graph-based framework, to improve AMR gene prediction by integrating gene neighbourhood information and protein-protein interaction networks. Focusing on Klebsiella pneumoniae and Escherichia coli, we construct a comprehensive and biologically grounded knowledge graph using curated data from CARD, STRING, and DrugBank. We apply knowledge graph embedding models which are fed into deep neural networks to infer novel AMR links, achieving classification accuracy of up to 97%. Our results demonstrate that incorporating biologically meaningful relationships, such as gene neighbourhood information and protein interactions, enhances the predictive accuracy and interpretability of AMR link predictions. This work contributes to the development of scalable and data-integrated approaches for advancing antimicrobial resistance surveillance and drug discovery. BRIDGE implementation and data are available at https://github.com/GraphML-lab/BRIDGE.

bioinformatics↗

bioRxiv2026

An endoplasmic reticulum resident molecular chaperone, GRP170, prevents stress-induced glomerular injury

The glomerulus, a unique capillary network in the nephron, filters an entire blood volume approximately 300 times a day. Specialized epithelial cells known as podocytes form a critical component of the glomerular filtration barrier, and diseases linked to podocyte injury include minimal change disease, focal segmental glomerulosclerosis, and diabetic kidney disease. Because podocytes are terminally differentiated, their ability to respond to external stress is critical. The unfolded protein response (UPR), a cellular stress pathway, is associated with glomerular injury, although the role of the UPR in glomerular injury is undefined. The UPR is initially protective, leading to upregulation of molecular chaperones, a class of proteins that promote protein folding and are required to survive oxidative and ischemic injury. An unresolved UPR, however, leads to apoptosis. We previously found that one molecular chaperone, GRP170, provides protection against acute kidney injury since GRP170 depletion led to UPR induction and widespread kidney injury. Here we generated a new podocyte specific GRP170 knock out mouse (GRP170Pd-/-). Surprisingly, GRP170Pd-/- mice were born healthy, and podocyte development appeared normal. Within a month, however, the knockout mice exhibited profound glomerular injury manifesting as proteinuria, hypoalbuminemia, hyperlipidemia, and kidney injury. Concomitant with glomerular injury, we observed increased expression of the pro-apoptotic UPR target, CHOP, in podocytes. Together, our new model not only defines a protective role for GRP170 against glomerular injury but also provides a new model to test the therapeutic potential of small molecule UPR modulators to treat glomerular injury.

physiology↗

bioRxiv2025

Mutations in the HKD motif of Trypanosoma brucei cardiolipin synthase inhibit cardiolipin synthesis and parasite growth

Phospholipases D (PLDs) are ubiquitous enzymes of the PLD superfamily that catalyze phosphodiester bond cleavage and transphosphatidylation reactions. They are characterized by the presence of two conserved HXK(X)4D(X)6G(X)2N or HKD motifs. While these motifs are essential for catalysis in diverse PLD family members, their functional significance in the bacterial-type cardiolipin synthase of the protozoan parasite Trypanosoma brucei (TbCls) has not yet been investigated. TbCls is essential for parasite survival in culture and contains two conserved HKD motifs in its amino acid sequence. Here we take advantage of a previously constructed cell line in which the endogenous alleles of TbCls were replaced with an inducible ectopic copy of the enzyme, to introduce point mutations in the HKD motifs of TbCls. We found that the HKD motif and its surrounding amino acids are essential for the de-novo synthesis of cardiolipin and for growth of the parasite.

biochemistry↗

bioRxiv2025

Protease secreting Nannochloropsis limnetica can valorise invasive plants through circular water-use and accumulate neutral lipids

AO_SCPLOWBSTRACTC_SCPLOWNannochloropsis limnetica can be cultivated on a wide range of effluents and generate high value lipids. Using a fed-batch system, this microalgae was used to demonstrate circular reuse of water to valorise protein depleted fractions (PDF) generated from Gorse leaf protein concentrate (LPC) production. Protein, phenolics and sugar content in the PDF was measured over seven days and corresponding microalgal lipid content was profiled. Extracellular protease activity was also screened to understand how N. limnetica digests and utilises proteins in the medium. A signal peptide-based approach predicted seven proteases expected to be secreted. N. limnetica could rapidly deplete nutrients (ksugar=0.20 {+/-} 0.06 day-1 and kproteins=0.17 {+/-} 0.04 day-1) from the wastewater which could be reused for subsequent LPC extraction processes without any impact on recovery or purity. Phenolics levels remained unaffected during microalgal growth and accumulated in the water across each extraction cycle. Microalgal neutral lipid content ([~]132.8 mglipid g-1 ) was higher in the effluent compared to standard BBM ([~]114.5 mglipid g-1 ). Casein agar plates and 1D zymogram confirmed extracellular protease production by N. limnetica, although their identities remains unclear owing to conflicting / fragmented data from the LC/MS analysis. This work shows that N. limnetica is able to secrete proteases to digest and utilise residual proteins in the medium, and can be employed to valorise agricultural wastewater. Identification of the proteases will require further investigation. HighlightsO_LIN. limnetica can rapidly deplete residual nutrients from LPC effluents. C_LIO_LIN. limnetica accumulates higher neutral lipids with LPC effluents versus BBM. C_LIO_LIN. limnetica secretes extracellular proteases to digest proteins in the medium. C_LIO_LIWater can be circularly reused between LPC production and N. limnetica cultivation. C_LI Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=108 SRC="FIGDIR/small/691304v2_ufig1.gif" ALT="Figure 1"> View larger version (34K): org.highwire.dtl.DTLVardef@1524de0org.highwire.dtl.DTLVardef@76cea1org.highwire.dtl.DTLVardef@1f7ad4aorg.highwire.dtl.DTLVardef@191f929_HPS_FORMAT_FIGEXP M_FIG C_FIG

microbiology↗

bioRxiv2025

Targeting tumor antioxidant pathways with novel GSTP1/M2 inhibitors for cancer treatment

Glutathione transferase (GSTP1 and GSTM2) are tractable targets for anticancer drug development. In this work, a series of 6-(7-nitro-2,1,3-benzoxadiazol-4-ylthio)hexanol (NBDHEX) based analogues were designed, synthesized, and evaluated both theoretically and experimentally as GSTP1 and GSTM2 inhibitors. Among the synthesized compounds, 3h showed selective inhibition toward GSTP1 while 5b showed selective inhibition of GSTM2. Compounds 5b and 5c exhibited stronger potency while compound 3h showed slightly lower potency against the tested cancer cells than its parent molecule NBDHEX. Comprehensive biological studies were conducted on the effect of 3h, 5b and 5c towards breast cancer MDA-MB-231 and pancreatic MiaPaCa-2 cell lines revealed that 3h, 5b and 5c could activate JNK pathway and induce cell apoptosis. Furthermore in vivo experiments using NSG mice demonstrated that 5b significantly reduced tumor growth when administered in combination with gemcitabine, effectively overcoming gemcitabine resistance in the MiaPaCa-2 cell model through targeted inhibition of GSTM2. These findings suggests that, 5b could become a promising candidate for further development as a potential antitumor agent in cancer therapy.

cancer biology↗

bioRxiv2025

Suppressing cortical glutamatergic neurons produces paradoxical interictal discharges and seizures

IntroductionSeizures are traditionally attributed to excessive excitation or deficient inhibition, yet recent clinical and slice data show they can also paradoxically arise when inhibition outweighs excitation. We chemogenetically suppressed neocortical glutamatergic neurons to test whether such suppression elicits epileptic activity in vivo. MethodsCaMKII-driven Gi-coupled hM4Di or Gq-coupled hM3Dq Designer Receptor Exclusively Activated by Designer Drug (DREADDs) were expressed in cortical glutamatergic neurons of Rasgrf2-jGCaMP8m and wildtype C57BL/6J mice. Widefield calcium imaging and 32-channel transparent electrocorticography were performed before and after systemic clozapine-N-oxide (CNO; 1.25-5 mg kg {superscript 1}). ResultsDREADD activation induced intermittent, large-amplitude, synchronized calcium transients confined to the CaMKII-hM4Di focus, accompanied by focal ECoG spikes that persisted for >3 h and sometimes evolved into seizures. By contrast, CNO activation of excitatory neurons via CaMKII-hM3Dq DREADDs desynchronized activity without epileptiform discharges. The same process duplicated in wild-type animals provoked similar epileptiform discharges. ConclusionSelective suppression of excitation can paradoxically drive cortical networks into hypersynchronous, epileptic states, challenging the simple excess-excitation model of ictogenesis. These findings highlight that seizures may result from relative inhibitory dominance, and that considering this excess inhibition mechanism could inspire new therapeutic approaches.

neuroscience↗

bioRxiv2025

Cell cycle-dependent variation in endocytosis drives phenotypic diversity in M. tuberculosis

Cell-to-cell heterogeneity is a hallmark of biology, yet how host variability shapes intracellular pathogen phenotypes is unclear. Using single cell approaches and redox-sensitive Mycobacterium tuberculosis reporters, we reveal that interphase-driven shifts in endocytic capacity create distinct intracellular niches. Bacilli residing in high-endocytic G2 cells adopt more oxidized redox states, whereas those in low-endocytic G1 cells remain reduced, generating phenotypic diversity within a single infected population. Experimental manipulation of host cell cycle stage reprograms endocytic capacity and intrabacterial redox, establishing a causal link between host state and pathogen diversification. Notably, the finding that the cell cycle regulates endocytic capacity constitutes a fundamental cell-biological discovery with broad implications. This variability persists in post-differentiated macrophages, indicating that proliferative history imprints functional diversity onto innate immune cells. Together, these results identify interphase-regulated endocytosis as a host-intrinsic mechanism that shapes Mycobacterium tuberculosis phenotypes and suggest new host-directed avenues to influence infection trajectories and persistence.

cell biology↗

bioRxiv2025

Cardiac Adaptations in the Cave Nectar Bat Eonycteris spelaea: Insights into Metabolic Resilience and Stress Response

AimsBats are unique mammals with remarkable adaptations, including powered flight, which demands significant energy expenditure. While previous studies have documented basic structural characteristics of bat hearts, a comprehensive understanding of their response to physiological stress remains unexplored. This study investigates the cardiac adaptations of the cave nectar bat Eonycteris spelaea to elucidate the mechanisms underlying their unique physiological capabilities. Methods and ResultsWe performed RNA sequencing to analyse the cardiac gene expression profile of E. spelaea and 6 other bat species in comparison to mouse and human hearts, revealing enriched transcriptomic signatures related to oxidative phosphorylation and fatty acid metabolism across multiple bat species. Metabolomic profiling compared acylcarnitine metabolites and tricarboxylic acid (TCA) cycle intermediates between bat and mouse hearts, indicating a distinct acylcarnitine profile and increased levels of TCA cycle intermediates in bats, suggesting enhanced metabolic capacity. Structural adaptations were assessed through anatomical and histological analyses on cardiac tissues, showing thicker left ventricular walls and increased vascular density in bats without pathological hypertrophy. Functional characteristics were evaluated using dobutamine stress echocardiography, demonstrating superior cardiac reserve in bats with significant increases in ejection fraction, stroke volume, and cardiac output under stress conditions. Additionally, isolated cardiomyocytes were treated with Angiotensin II (Ang II) to assess stress responses. Bat cardiomyocytes displayed resistance to Ang II-induced hypertrophy and mitochondrial dysfunction compared to mice, further highlighting their resilience to stress-induced damage. ConclusionThe unique adaptations observed in bat hearts, including enhanced metabolic pathways, structural remodelling, and cellular resilience contribute to their ability to meet the high energy demands of powered flight while maintain cardiac function under stress. These insights into bat cardiac physiology provide valuable information on cardioprotective mechanisms that could be applicable to other species. Translational PerspectiveBats exhibit remarkable cardiac adaptations that sustain the high energy demands of powered flight while resisting stress-induced damage. These insights into evolutionarily conserved cardioprotective mechanisms highlight potential therapeutic pathways for preventing heart failure, including resistance to hypertrophy and mitochondrial dysfunction under stress. Studying non-model organisms like bats may offer innovative approaches to enhance metabolic and stress resilience in human hearts, paving the way for translational research in cardiovascular disease management and treatment.

systems biology↗

bioRxiv2025

Enhanced delivery of lipid nanoparticle-based immunotherapy by modulating the tumor tissue stiffness using ultrasound-activated nanobubbles

Tumors often exhibit an extracellular matrix with elevated stiffness due to excessive accumulation and crosslinking of proteins, particularly collagen. This elevated stiffness acts as a physical barrier, impeding the infiltration of immune cells and the effective delivery of various immunotherapeutic agents, such as lipid nanoparticle-based RNA therapeutics. Here, we investigate the ability of ultrasound-activated nanobubbles (US-NBs) to increase the permeability and immunogenicity of tumors. Our results show that US-NBs physically remodel the tumor tissue by decreasing its stiffness by 60% five days after a single treatment. US-NB-treated tumors display randomly oriented collagen with a 5.47-fold lower deposition compared to untreated tumors. This leads to the effective delivery and widespread distribution of lipid nanoparticles (LNPs) in the tumor. When LNPs are assisted by US-NB, they have higher gene-transfection across pan-immune cells relative to LNPs alone. Notably, US-NB enables LNPs to genetically modify T cells directly in vivo. By effectively engaging both arms of the immune system, US-NB-assisted LNPs enhance the tumor immunogenicity and infiltration of cytotoxic cells by 4-fold when compared to LNPs alone. These results indicate that gentle mechanical stimulation of the tumor using US-NB offers a promising strategy to augment the delivery and efficacy of existing immunotherapies. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=142 SRC="FIGDIR/small/648512v1_ufig1.gif" ALT="Figure 1"> View larger version (48K): org.highwire.dtl.DTLVardef@da7fborg.highwire.dtl.DTLVardef@8f5f27org.highwire.dtl.DTLVardef@b002f6org.highwire.dtl.DTLVardef@19265da_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioengineering↗

bioRxiv2025

Transcription factor LHX2 suppresses astrocyte proliferation in the postnatal mammalian cerebral cortex

In the developing cerebral cortex astrocytes arise from progenitors in the ventricular and subventricular zones (V-SVZ), and also from local proliferation within the parenchyma. In the mouse neocortex, astrocytes that occupy upper versus deep layers (UL/DL) are known to be distinct populations in terms of molecular and morphological features. Transcription factor LHX2 is expressed both in V-SVZ gliogenic progenitors and in differentiated astrocytes throughout development and into adulthood. Here we show that loss of Lhx2 at birth results in an increased astrocyte proliferation in UL but not the DL of the cortex in the first postnatal week. Consistent with this, transcriptomic signatures of UL astrocytes increase. By 3 months, Lhx2 mutant astrocytes display upregulation of GFAP, and transcriptomic signatures associated with "reactive" astrocytes, in the absence of injury. These results demonstrate a novel role for Lhx2 in regulating proliferation and molecular features of cortical astrocytes. Summary StatementLoss of Lhx2 causes increased astrocyte proliferation in the upper but not deep layers of the mammalian cortex, and upregulation of reactive gliosis-like signatures in the absence of injury.

neuroscience↗

bioRxiv2025

Using network component analysis to study axenisation strategies for phototrophic eukaryotic microalgae

BackgroundThe axenisation of phototrophic eukaryotic microalgae has been studied for over a century, with antibiotics commonly employed to achieve axenic cultures. However, this approach often yields inconsistent outcomes and may contribute to the emergence of antibiotic-resistant microbes. A comprehensive review of microalgal species and the methods used to achieve axeny could provide insights into potentially effective workflows and identify gaps for future exploration. MethodsScholarly databases were systematically searched, supplemented by citation network analysis and AI-assisted tools, to collect studies on achieving axenic phototrophic eukaryotic microalgae cultures. Data about microalgal species, axenisation workflows, outcomes, and related factors (e.g., sampling locations, axenisation confirmation methods) were summarised. Network component analysis was used to identify clusters of commonly reported methods for diatoms, dinoflagellates, and green algae. A scoring framework was developed to assess the quality and reliability of evidence presented in the studies. ResultsPromising workflows circumventing the use of antibiotics appear to be filtration {leftrightarrow} washing {leftrightarrow} micropicking for diatoms, micropicking {leftrightarrow} subculturing {leftrightarrow} flow cytometry for dinoflagellates, and anoxy {leftrightarrow} photosensitisation {leftrightarrow} streak plating for green algae. Evidence from the literature indicates that a combination of microscopy (e.g., epifluorescence), cell counting (e.g., agar plating), and sequencing (16S and/or 18S) could enhance confidence in confirming axeny. ConclusionMore systematic and high quality primary research is required to identify effective workflows for other microalgal divisions and fortify / contradict the ones proposed herein based on network component analysis.

microbiology↗

bioRxiv2025

Proximity-labeling proteomics reveals remodeled interactomes and altered localization of pathogenic SHP2 variants

Missense mutations in PTPN11, which encodes the protein tyrosine phosphatase SHP2, are common in several developmental disorders and cancers. While many mutations disrupt auto-inhibition and hyperactivate SHP2, several do not enhance catalytic activity. Both activating and non-activating mutations could potentially drive pathogenic signaling by altering SHP2 interactions or localization. We employed proximity-labeling proteomics to map the interaction networks of wild-type SHP2, ten clinically-relevant mutants, and SHP2 bound to an inhibitor that stabilizes its auto-inhibited state. Our analyses revealed mutation- and inhibitor-dependent alterations in the SHP2 interactome, with several mutations also changing localization. Some mutants had increased mitochondrial localization and impacted mitochondrial function. This study provides a resource for exploring SHP2 signaling and offers new insights into the molecular basis of SHP2-driven diseases. Furthermore, this work highlights the capacity for proximity-labeling proteomics to detect missense-mutation-dependent changes in protein interactions and localization.

molecular biology↗

bioRxiv2025

Artificial Intelligence Apps for Medical Image Analysis using pyCERR and Cancer Genomics Cloud

This work introduces a user-friendly, cloud-based software framework for conducting Artificial Intelligence (AI) analyses of medical images. The framework allows users to deploy AI-based workflows by customizing software and hardware dependencies. The components of our software framework include the Python-native Computational Environment for Radiological Research (pyCERR) platform for radiological image processing, Cancer Genomics Cloud (CGC) for accessing hardware resources and user management utilities for accessing images from data repositories and installing AI models and their dependencies. GNU-GPL copyright pyCERR was ported to Python from MATLAB-based CERR to enable researchers to organize, access, and transform metadata from high dimensional, multi-modal datasets to build cloud-compatible workflows for AI modeling in radiation therapy and medical image analysis. pyCERR provides an extensible data structure to accommodate metadata from commonly used medical imaging file formats and a viewer to allow for multi-modal visualization. Analysis modules are provided to facilitate cloud-compatible AI-based workflows for image segmentation, radiomics, DCE MRI analysis, radiotherapy dose-volume histogram-based features, and normal tissue complication and tumor control models for radiotherapy. Image processing utilities are provided to help train and infer convolutional neural network-based models for image segmentation, registration and transformation. The framework allows for round-trip analysis of imaging data, enabling users to apply AI models to their images on CGC and retrieve and review results on their local machine without requiring local installation of specialized software or GPU hardware. The deployed AI models can be accessed using APIs provided by CGC, enabling their use in a variety of programming languages. In summary, the presented framework facilitates end-to-end radiological image analysis and reproducible research, including pulling data from sources, training or inferring from an AI model, utilities for data management, visualization, and simplified access to image metadata.

bioinformatics↗

bioRxiv2024

Ilastik: a machine learning image analysis platform to interrogate stem cell fate decisions across multiple vertebrate species

Stem cells are the key cellular source for regenerating tissues and organs in vertebrate species. Historically, the investigation of stem cell fate decisions in vivo has been assessed in tissue sections using immunohistochemistry (IHC), where a trained user quantifies fluorescent signal in multiple randomly selected images using manual counting--which is prone to inaccuracies, bias, and is very labor intensive. Here, we highlight the performance of a recently developed machine-learning (ML)-based image analysis program called Ilastik using skeletal muscle as a model system. Interestingly, we demonstrate that Ilastik accurately quantifies Paired Box Protein 7 (PAX7)-positive muscle stem cells (MuSCs) before and during the regenerative process in whole muscle sections from mice, humans, axolotl salamanders, and short-lived African turquoise killifish, to a precision that exceeds human capabilities and in a fraction of the time. Overall, Ilastik is a free user-friendly ML-based program that will expedite the analysis of stained tissue sections in vertebrate animals.

cell biology↗

bioRxiv2024

High-Throughput Targeted Drug Screening for NF1-associated High-Grade Gliomas with ATRX Deficiency

Neurofibromatosis type 1 (NF1)-associated high-grade gliomas (HGGs) harboring ATRX mutations exhibit an aggressive clinical phenotype, driven by heightened genomic instability and metabolic reprogramming. Existing therapies, including chemotherapy and radiotherapy, are limited by resistance mechanisms and formation of secondary malignancy, underscoring the need for novel therapeutic strategies. Here, we report the results of a high-throughput screening of 10,000 small molecules aimed at identifying compounds selectively targeting vulnerabilities associated with concurrent ATRX and NF1 loss. Among the screened compounds, K784-6195 (ChemDiv ID) emerged as a promising candidate, exhibiting marked selective cytotoxicity in NF1-associated glioma cell lines with ATRX deficiency (IC50 = 4.84 {micro}M). In comparison, wild-type ATRX sporadic glioma cell lines (U251) exhibited significantly reduced sensitivity to K784-6195 (IC50 = 37.03 {micro}M). However, ATRX knockout U251 glioma cells recapitulating concurrent ATRX and NF1 loss exhibited heightened susceptibility to K784-6195 (IC50 = 20-23 {micro}M) compared to their wild-type counterpart. Metabolomic analysis revealed that K784-6195 treatment impairs metabolic pathways, including the pentose phosphate pathway, glutamine metabolism, and redox homeostasis, leading to oxidative stress and impaired cell survival. These findings highlight K784-6195 as a promising candidate for therapeutic development, offering a targeted approach for the treatment of NF-1 associated HGGs with ATRX deficiency.

cancer biology↗