Biology subjects

Khan, S.

Publications and source records attributed to Khan, S..

At least 91 records · Page 5Linked to original sources

bioRxiv2024

An opioid efficacy switch for reversible optical control of peripheral analgesia

The mu-opioid receptor (MOR) is a major target for the treatment of pain. However, opioids are prone to side effects which limit their effectiveness as analgesics and can lead to opioid use disorders or, even, lethal overdose. The systemic administration of opioid agonists makes it both very difficult to decipher their underlying circuit mechanisms of action and to limit drug action to specific receptor subpopulations to isolate therapeutic effects from adverse side effects. Here we design, synthesize, and characterize a reversibly photoswitchable morphinan agonist termed "azo-morphine-3" (AM-3) which interconverts from low to high efficacy in response to different wavelengths of light to enable optical control of MOR signaling. Cryo-EM structures of the low efficacy "trans" and high efficacy "cis" states of AM-3 bound to the MOR reveal distinct binding modes of the photoswitchable azobenzene moiety, each inducing unique structural dynamics, providing insight into the molecular basis of agonist efficacy. In mice, AM-3 drives reversible and repeatable optical control of anti-nociception with a reduced side effect profile owing to its restriction to the periphery and its ability to be locally activated at the site of pain.

pharmacology and toxicology↗

bioRxiv2024

Genomic Foundationless Models: Pretraining Does Not Promise Performance

AO_SCPLOWBSTRACTC_SCPLOWThe success of Large Language Models has inspired the development of Genomic Foundation Models (GFMs) through similar pretraining techniques. However, the relationship between pretraining performance and effectiveness in downstream genomic tasks remains unclear. Additionally, the high computational cost of pretraining raises questions about its cost-efficiency. To assess the usefulness of pretraining in genomics, we evaluated seven different GFMs across 52 diverse genomic tasks, comparing them to their counterparts with randomly initialized weights. Across benchmarks, we find that randomly initialized models provide surprisingly strong baselines and tokenizer and architecture choices strongly shape both these baselines and the gains from pretraining. Specifically, character-token models often match or exceed the performance of larger pretrained k-mer or BPE models, whereas subword models appear to benefit from pretraining. We also find that the evaluated GFMs fail to capture clinically relevant genetic mutations, with embeddings and log-likelihood ratios showing limited sensitivity to annotated variants. For the tasks we study, these results suggest that current NLP-style pretraining strategies provide modest, tokenizer-gated improvements over strong random baselines and motivate more biologically informed tokenization and variant-aware objectives. Our code is available at github.com/m42-health/gfm-random-eval.

genomics↗

bioRxiv2024

Pan-cancer N-glycoproteomic atlas of patient derived xenografts uncovers FAT2 as a therapeutic target for head and neck cancers

Cell surface proteins offer significant cancer therapeutic potential attributable to their accessible membrane localization and central role in cellular signaling. Despite this, their promise remains largely untapped due to the technical challenges inherent to profiling cell surface proteins. Here, we employed N-glycoproteomics to analyze 85 patient-derived xenografts (PDX), constructing Glyco PDXplorer - an in vivo pan-cancer atlas of cancer-derived cell surface proteins. We developed a target discovery pipeline to prioritize proteins with favorable expression profiles for immunotherapeutic targeting and validated FAT2 as a head and neck squamous cancer (HNSC) enriched surface protein with limited expression in normal tissue. Functional studies revealed that FAT2 is essential for HNSC growth and adhesion through regulation of surface architecture and integrin-PI3K signaling. Chimeric antigen receptor (CAR) T cells targeting FAT2 demonstrated potent anti-tumor activity in HNSC models. This work lays the foundation for developing FAT2-targeted therapies and represents a pivotal resource to inform therapeutic target discovery for multiple cancers. HIGHLIGHTSO_LIPan-cancer landscape of cancer-derived cell surface proteins detected in vivo C_LIO_LIDevelopment of a multi-omic discovery pipeline to prioritize proteins with optimal expression profiles as immunotherapy targets C_LIO_LIIdentification and validation of FAT2 as a head and neck squamous cancer enriched surface protein with minimal expression in normal tissues C_LIO_LIFAT2 coordinates cell surface organization, adhesion, growth and survival through the integrin-PI3K-AKT pathway C_LIO_LIFAT2 CAR T cells demonstrate anti-tumour activity in pre-clinical models C_LI

cancer biology↗

bioRxiv2024

Calibration-free estimation of field dependent aberrations for single molecule localization microscopy across large fields of view

Image quality in single molecule localization microscopy (SMLM) depends largely on the accuracy and precision of the localizations. While under ideal imaging conditions the theoretically obtainable precision and accuracy are achieved, in practice this changes if (field dependent) aberrations are present. Currently there is no simple way to measure and incorporate these aberrations into the Point Spread Function (PSF) fitting, therefore the aberrations are often taken constant or neglected all together. Here we introduce a model-based approach to estimate the field-dependent aberration directly from single molecule data without a calibration step. This is made possible by using nodal aberration theory to incorporate the field-dependency of aberrations into our fully vectorial PSF model. This results in a limited set of aberration fit parameters that can be extracted from the raw frames without a bead calibration measurement, also in retrospect. The software implementation is computationally efficient, enabling fitting of a full 2D or 3D dataset within a few minutes. We demonstrate our method on 2D and 3D localization data of microtubuli and nuclear pore complexes over fields of view (FOV) of up to 180 m and compare it with spline-based fitting and a deep learning based approach.

microbiology↗

bioRxiv2024

Pharmacological inhibition of 11βhydroxysteroid dehydrogenase type 1 after myocardial infarction targets extracellular matrix processing and preserves cardiac function in a translational mini-pig model

Background and PurposePlasma glucocorticoids (GCs) increase acutely after myocardial infarction (MI), thereafter tissue levels are amplified selectively within cells expressing 11{beta}Hydroxysteroid Dehydrogenase type 1 (11{beta}HSD1) that regenerates active GCs from circulating metabolites. GCs initially protect cardiomyocytes and prevent excessive inflammation after MI but can also suppress subsequent wound repair leading to functional decline. The present study aimed to investigate the potential of pharmacological 11{beta}HSD1 inhibition after MI to prevent deterioration of cardiac function and its impact on wound repair. Experimental ApproachAdult female Gottingen mini-pigs underwent percutaneous balloon MI/reperfusion and were randomised to receive either oral 11{beta}HSD1 inhibitor AZD8329 (n=11), or vehicle (n=9), from 2 until 27 days later, with concurrent administration of clinically relevant therapeutic intervention (anti-platelet, statin and ACE inhibitor). Key ResultsAZD8329 treatment increased plasma accumulation of cortisone substrate consistent with successful 11{beta}HSD1 inhibition. Gadolinium-enhanced MRI showed equivalent infarct size in both groups prior to commencing treatment. 28 days after MI cardiac function and LV area were preserved in the AZD8329 treated group relative to vehicle. There was no impact of 11{beta}HSD1 inhibitor on neovascularisation or infarct area. Mass spectrometry imaging revealed AZD8329 binding to the healing infarct and altered regulation of extracellular matrix (ECM) processing was highlighted by birefringence microscopy and proteomic analysis. Conclusions and ImplicationsPharmacological inhibition of 11{beta}HSD1 after MI prevents deterioration of cardiac function and detrimental remodelling. 11{beta}HSD1 inhibitors have safely reached phase 2 clinical trials in diabetes and dementia and could be repurposed as an addition to standard care after MI to prevent the development of heart failure. Bullet Point Summary What is already known?O_LIGCs are released from the adrenal gland after MI, but also regenerated within the heart from circulating precursors by the enzyme 11{beta}HSD1 in cardiomyocytes, fibroblasts and macrophages. C_LIO_LIGenetic suppression of Hsd11b1 expression in the mouse promotes neovascularisation, prevents infarct expansion during infarct repair after MI and the development of heart failure. C_LI What does this study add?O_LIOral pharmacological inhibition of 11{beta}HSD1 after MI/reperfusion in a translational mini-pig model of MI receiving concurrent clinically relevant therapy prevents cardiac functional deterioration and adverse ventricular remodelling over the following 4 weeks. C_LIO_LIMass spectrometry imaging reveals target engagement of the 11{beta}HSD1i in the repairing infarct. C_LIO_LIThe mechanism is independent of neovascularisation but does involve modification of extracellular matrix remodelling during repair and scar formation. C_LI What is the clinical significance?O_LITissue 11{beta}HSD1 expression is increased in aging when the risk of MI is higher. C_LIO_LIPharmacological inhibitors of 11{beta}HSD1 have safely reached phase 2 clinical trials for dementia and metabolic disease and could be repurposed for use post-MI to prevent the development of heart failure. C_LI

pharmacology and toxicology↗

bioRxiv2024

Recommendations for Bioinformatics in Clinical Practice

Next Generation Sequencing (NGS) is increasingly used in clinical diagnostics, largely driven by the success and robustness of Whole Genome Sequencing (WGS). Whereas updated guidelines exist for how to interpret and report on variants that are identified from NGS using bioinformatics pipelines, there is a need for standardised bioinformatics practices for diagnostics to ensure clinical consensus, accuracy, reproducibility and comparability of the results. This article presents consensus recommendations developed by 13 clinical bioinformatics units taking part in the Nordic Alliance for Clinical Genomics (NACG), by expert bioinformaticians working in clinical production. The recommendations are based on clinical practice and focus on analysis types, test and validation, standardisation and accreditation, as well as core competencies and technical management required for clinical bioinformatics operations. Key recommendations include adopting the hg38 genome build as the reference and a standard set of recommended analyses, including the use of multiple tools for structural variant (SV) calling and in-house data sets for filtering recurrent calls. Clinical bioinformatics production should operate under the ISO 15189 standard, utilising off-grid clinical-grade high-performance computing systems, standardised file formats, and strict code version control. Containerized software containers or environment management systems are needed to ensure reproducibility. Pipelines should be rigorously documented and tested for accuracy and reproducibility, minimally covering unit, integration, and end-to-end testing. Standard truth sets such as GIAB and SEQC2 for germline and somatic variant calling, respectively, should be supplemented by recall testing of previously validated clinical cases. Data integrity must be verified using file hashing, and sample identity should be checked via sample fingerprinting and genetically inferred identification markers such as sex and relatedness. Finally, clinical bioinformatics teams should encompass diverse skills, including software development, data management, quality assurance, and domain expertise in human genetics. These recommendations provide a consensus framework for standardising bioinformatics practices across clinical WGS applications and can serve as a practical guide to facilities that are new to large-scale sequencing-based diagnostics, or as a reference for those who already run high-volume clinical production using NGS.

bioinformatics↗

bioRxiv2024

Axonal injury signaling is restrained by a spared synaptic branch

The intrinsic ability of injured neurons to degenerate and regenerate their axons facilitates nervous system repair, however this ability is not engaged in all neurons and injury locations. Here we investigate the regulation of a conserved axonal injury response pathway with respect to the location of damage in branched motoneuron axons in Drosophila larvae. The dileucine zipper kinase DLK, (also known as MAP3K12 in mammals and Wallenda (Wnd) in Drosophila), is a key regulator of diverse responses to axonal injury. In three different populations of motoneurons, we observed the same striking result that Wnd/DLK signaling becomes activated only in response to injuries that remove all synaptic terminals. Injuries that spare even a small part of a synaptic terminal fail to activate Wnd/DLK signaling, despite the presence of extensive axonal degeneration. The regulation of injury-induced Wnd/DLK signaling occurs independently of its previously known regulator, the Hiw/PHR ubiquitin ligase. We propose that Wnd/DLK signaling regulation is linked to the trafficking of a synapse-to-nucleus axonal cargo and that this mechanism enables neurons to respond to impairments in synaptic connectivity.

neuroscience↗

bioRxiv2024

Mu opioid receptor expression by nucleus accumbens inhibitory interneurons promotes affiliative social behavior

Mu opioid receptors in the nucleus accumbens regulate motivated behavior, including pursuit of natural rewards like social interaction as well as exogenous opioids. We used a suite of genetic and viral strategies to conditionally delete mu opioid receptor expression from all major neuron types in the nucleus accumbens. We pinpoint inhibitory interneurons as an essential site of mu opioid receptor expression for typical social behavior, independent from exogenous opioid sensitivity.

neuroscience↗

bioRxiv2024

Inferring post-transcriptional regulation within and across cell types in human testis

Single-cell tissue atlases commonly use RNA abundances as surrogates for protein abundances. Yet, protein abundance also depends on the regulation of protein synthesis and degradation rates. To estimate the contributions of such post transcriptional regulation, we quantified the proteomes of 5,883 single cells from human testis using 3 distinct mass spectrometry methods (SCoPE2, pSCoPE, and plexDIA). To distinguish between biological and technical factors contributing to differences between protein and RNA levels, we developed BayesPG, a Bayesian model of transcript and protein abundance that systematically accounts for technical variation and infers biological differences. We use BayesPG to jointly model RNA and protein data collected from 29,709 single cells across different methods and datasets. BayesPG estimated consensus mRNA and protein levels for 3,861 gene products and quantified the relative protein-to-mRNA ratio (rPTR) for each gene across six distinct cell types in samples from human testis. About 28% of the gene products exhibited significant differences at protein and RNA levels and contributed to about 1, 500 significant GO groups. We observe that specialized and context specific functions, such as those related to spermatogenesis are regulated after transcription. Among hundreds of detected post translationally modified peptides, many show significant abundance differences across cell types. Furthermore, some phosphorylated peptides covary with kinases in a cell-type dependent manner, suggesting cell-type specific regulation. Our results demonstrate the potential of inferring protein regulation in from single-cell proteogenomic data and provide a generalizable model, BayesPG, for performing such analyses.

systems biology↗

bioRxiv2024

A clinical mutation in uvrA, a DNA repair gene, confers survival advantage to Mycobacterium tuberculosis in the host

DNA repair pathways play an essential role in maintaining the genomic integrity of bacteria, and a perturbation in their biological activity helps bacteria survive under duress. In drug-resistant clinical strains, we identified a Q135K mutation in the uvrA gene, a DNA repair pathway gene. To delineate the role of uvrA and the Q135K mutation, we generated the gene replacement mutant of UvrA (Rv{Delta}uvrA) in Mycobacterium tuberculosis H37Rv (Mtb-Rv). While the lack of UvrA function in Rv{Delta}uvrA could be restored upon complementation with uvrA, the uvrA-Q135K mutant identified in clinical drug-resistant strains failed to do so. This was reflected in higher mutation rates in Rv{Delta}uvrA and Rv{Delta}uvrA::uvrAQ135A, compared with wild-type Rv or Rv{Delta}uvrA::uvrA complemented strains in the presence and absence of oxidative stress. Killing kinetics experiments with anti-TB drugs showed increased survival of Rv{Delta}uvrA and Rv{Delta}uvrA::uvrAQ135K, strains compared with Rv or Rv{Delta}uvrA::uvrA. Importantly, Rv{Delta}uvrA and Rv{Delta}uvrA::uvrAQ135K showed enhanced survival in peritoneal macrophages and murine infection model of infection. Together, data suggests that acquiring Q135K mutation benefits the pathogen, which helps enhance the hosts survival adaptability. Author SummaryDNA repair mechanisms in an organism are necessary for correcting the errors generated during replication or when it is damaged/modified due to insults. As a GC organism, Mtb is highly prone to host-mediated attacks on its genome, which, if uncorrected, can impact its genome integrity. The drug-resistant clinical strains of Mtb harbor Q135K mutation in uvrA, the first enzyme in the nucleotide excision repair pathway. With the help of genetic, molecular, and murine challenge experiments, we show that the UvrA-Q135K mutation abrogates the enzymes activity, compromising the Mtb strain harboring the mutation in the oxidative and nitrosative stress. On the contrary, the mutation in UvrA imparts survival advantage in activated macrophages and murine infection models. Results presented argue that identified mutation helps in better adaptability in the host, which may include faster acquisition of drug resistance.

microbiology↗

bioRxiv2024

Efficacy of photodynamic therapy using 5-aminolevulinic acid-induced photosensitization is enhanced in pancreatic cancer cells with acquired drug resistance

The use of 5-aminolevulinic acid (ALA) as a precursor for protoporphyrin IX (PpIX) is an established photosensitization strategy for photodynamic therapy (PDT) and fluorescence guided surgery. Ongoing studies are focused on identifying approaches to enhance PpIX accumulation as well as to identify tumor sub-types associated with high PpIX accumulation. In this study, we investigated PpIX accumulation and PDT treatment response with respect to nodule size in 3D cultures of pancreatic cancer cells (Panc1) and a derivative subline (Panc1OR), which has acquired drug resistance and exhibits increased epithelial mesenchymal transition. In monolayer and 3D culture dose response studies the Panc1OR cells exhibit significantly higher cell killing at lower light doses than the drug naive cells. Panc1OR also exhibits increased PpIX accumulation. Further analysis of cell killing efficiency per molecule of intracellular PpIX indicates that the drug resistant cells are intrinsically more responsive to PDT. Additional investigation using exogenous delivery of PpIX also shows higher cell killing in drug resistant cells, under conditions which achieve approximately the same intracellular PpIX. Overall these results are significant as they demonstrate that this example of drug-resistant cells associated with aggressive disease progression and poor clinical outcomes, show increased sensitivity to ALA-PDT.

cancer biology↗

bioRxiv2024

Plant-Pollinator Interactions along the Altitudinal Gradient in Berberis lycium Royle: An Endangered Medicinal Plant Species of the Himalayan Region

Mountain ecosystems influence species distribution by offering climatic variables intertwined with rising altitude. These climatic factors determine species phenology and niche width. Although the distributional patterns of some prominent insect groups in relation to altitude have been determined, the environmental preferences along the altitudinal range that differentially influence the pollination of specific plant species are unknown. Here we assess how the composition and abundance of pollinator fauna of the important medicinal plant Berberis lycium Royle (Berberidaceae) differ across five distinct altitudinal gradients (800-2200 m) in the Pir-Panjal mountain range in the northwestern part of the Indian Himalayas. We monitored insect pollinators of major groups (bees, butterflies, wasps, flies) over two consecutive flowering seasons during 2022-2023. In total, 39 insect species belonging to five orders and 17 families were observed visiting the plant species during the flowering period across the altitudinal gradient. The results of the linear regression model depict that all four pollination indices show a negative correlation with increasing altitude in foraging activities when all the data are pooled together. However, only foraging speed (FS) and index of visiting rate (IVR) were statistically significant. In the case of individual orders, only Lepidoptera exhibited a notable relation to altitude. However, asynchrony in foraging activities among other pollinator groups has been reported along altitudinal gradients. The reproductive output (fruit and seed production) shows a significant negative correlation with increasing altitude. We concluded that while altitude influences species distribution, it also differentially shapes plant-pollinator interactions, pollinator activities, and reproductive output. This work is of great significance in order to monitor plant-pollinator interactions, which are essential component of biodiversity rich but fragile mountain ecosystem.

ecology↗

bioRxiv2024

Proteome asymmetry in mouse and human embryos before fate specification

Pre-patterning of the embryo, driven by spatially localized factors, is a common feature across several non-mammalian species1-4. However, mammals display regulative development and thus it was thought that blastomeres of the embryo do not show such pre-patterning, contributing randomly to the three lineages of the blastocyst: the epiblast, primitive endoderm and trophectoderm that will generate the new organism, the yolk sac and placenta respectively 4-6. Unexpectedly, early blastomeres of mouse and human embryos have been reported to have distinct developmental fates, potential and heterogeneous abundance of certain transcripts7-12. Nevertheless, the extent of the earliest intra-embryo differences remains unclear and controversial. Here, by utilizing multiplexed and label-free single-cell proteomics by mass-spectrometry13, we show that 2-cell mouse and human embryos contain an alpha and a beta blastomere as defined by differential abundance of hundreds of proteins exhibiting strong functional enrichment for protein synthesis, transport, and degradation. Such asymmetrically distributed proteins include Gps1 and Nedd8, depletion or overexpression of which in one blastomere of the 2-cell embryo impacts lineage segregation. These protein asymmetries increase at 4-cell stage. Intriguingly, halved mouse zygotes display asymmetric protein abundance that resembles alpha and beta blastomeres, suggesting differential proteome localization already within zygotes. We find that beta blastomeres give rise to a blastocyst with a higher proportion of epiblast cells than alpha blastomeres and that vegetal blastomeres, which are known to have a reduced developmental potential, are more likely to be alpha. Human 2-cell blastomeres also partition into two clusters sharing strong concordance with clusters found in mouse, in terms of differentially abundant proteins and functional enrichment. To our knowledge, this is the first demonstration of intra-zygotic and inter-blastomere proteomic asymmetry in mammals that has a role in lineage segregation.

developmental biology↗

bioRxiv2024

Dynamic convergence of autism disorder risk genes across neurodevelopment

Over three hundred and seventy-three risk genes, broadly enriched for roles in neuronal communication and gene expression regulation, underlie risk for autism spectrum disorder (ASD) and developmental delay (DD). Functional genomic studies of subsets of these genes consistently indicate a convergent role in neurogenesis, but how these diverse risk genes converge on a smaller number of biological pathways in mature neurons is unclear. To uncover shared downstream impacts between neurodevelopmental disorder (NDD) risk genes, here we apply a pooled CRISPR approach to contrast the transcriptomic impacts of targeting 29 NDD loss-of-function genes across human induced pluripotent stem cell (hiPSC)-derived neural progenitor cells, glutamatergic neurons, and GABAergic neurons. Points of convergence vary between the cell types of the brain and are greatest in mature glutamatergic neurons, where they broadly target not just synaptic and epigenetic, but unexpectedly, mitochondrial biology. The strongest convergent networks occur between NDD genes with common co-expression patterns in the post-mortem brain, biological annotations, and clinical associations, suggesting that convergence may one-day inform patient stratification and treatment. Towards this, ten out of eleven drugs tested that were predicted to reverse convergent signatures in human cells and/or arousal and sensory processing behaviors in zebrafish ameliorated at least one behavioral phenotype in vivo. Altogether, robust convergence in post-mitotic neurons represents a clinically actionable therapeutic window.

genetics↗

bioRxiv2024

Mitochondrial calcium uniporter regulates human fibroblast-like synoviocytes invasion via altering mitochondrial dynamics and dictates rheumatoid arthritis pathogenesis

Rheumatoid arthritis (RA) is characterized by the aggressive migration and invasion of fibroblast-like synoviocytes (FLS) into cartilage and bone, a process that is significantly influenced by mitochondrial calcium uptake. This study highlights the critical role of mitochondrial calcium uniporter (MCU) in regulating FLS migration and mitochondrial dynamics, and its potential as a therapeutic target in RA. Notably, RA-FLS exhibited increased MCU expression and mitochondrial dysfunction compared to controls. Treatment with Ru360, a potent MCU inhibitor significantly reduced RA-FLS migration, calcium influx, and mitochondrial reactive oxygen species (ROS) levels, while restoring mitochondrial morphology and enhancing ATP production. Further analysis of MCU complex expression revealed elevated levels of MCU and other regulatory subunits (EMRE, MICU1, MICU2) in RA-FLS compared to controls, indicating mitochondrial dysfunction in RA. Mechanistically, MCU inhibition altered gene expression related to cytoskeletal dynamics, focal adhesion, and metabolic pathways. In human RA-FLS, MCU inhibition suppressed migration and invasion both in vitro and in vivo. The interaction between MCU and Miro1, essential for mitochondrial transport, was validated, and its disruption impaired FLS migration. Our findings highlight MCU as a promising therapeutic target to inhibit FLS migration and ameliorate RA progression.

immunology↗

bioRxiv2024

Scalable and universal prediction of cellular phenotypes

Biological systems can be interrogated by perturbing individual components and observing the consequences across molecular, cellular, and phenotypic levels. The vast combinatorial space of possible perturbations and responses makes exhaustive experimentation infeasible. Recent advances in machine learning have shown that training on diverse datasets enables transfer learning across tasks, capturing patterns that generalize and improving performance on previously unseen problems. Inspired by this principle, we present Prophet, a transformer-based model pretrained on a vast, heterogeneous collection of perturbation experiments. This pretraining allows Prophet to predict the outcomes of untested genetic or chemical perturbations in novel cellular contexts, spanning phenotypes such as gene expression, viability, and morphology. By leveraging shared structure across apparently disconnected assays, Prophet provides a scalable framework for large-scale virtual screening and prioritization of informative experiments. Prophet consistently outperforms baseline models, including those trained on single phenotypes, showing that transfer learning between phenotypes not only is possible but improves predictive accuracy. Its capabilities extends to in vivo developmental systems, where it recapitulates known lineage biology and proposes new candidates. In a large-scale in silico screen for melanoma, Prophet identified and experimentally validated compounds with selective activity that mirrored clinically approved therapies, demonstrating its ability to transform perturbation biology into a predictive and scalable engine for therapeutic discovery.

cell biology↗

bioRxiv2024

PrePR-CT: Predicting Perturbation Responses in Unseen Cell Types Using Cell-Type-Specific Graphs

Predicting the transcriptional response of chemical perturbations is crucial to understanding gene function and developing drug candidates, promising a streamlined drug development process. Single-cell sequencing has provided an ideal data basis for training machine learning models for this task. Recent advances in deep learning have led to significant improvements in predictions of chemical as well as genetic perturbations at the single cell level. Experiments have shown that different cell types exhibit distinct transcriptional patterns and responses to perturbation. This poses a fundamental problem for predicting transcriptional responses of drugs or cell types outside the training data. Accordingly, existing methods lack cell-type-specific modeling or do not explicitly provide an interpretable mechanism for the gene features. In this study, we introduce a novel approach that employs a network representation of various cell types as an inductive bias, improving prediction performance in scenarios with limited data while acknowledging cellular differences. We applied our framework to four small-scale single-cell perturbation datasets and one large-scale screening experiment, demonstrating that this representation can inherently generalize to previously unseen cell types. Furthermore, our method outperforms the state-of-the-art methods in predicting the post-perturbation response in unobserved cell types.

bioinformatics↗

bioRxiv2024

Mineral Phase Changes During Intervertebral Disc Degeneration

Intervertebral disc disease is a common cause of pain and neurological deficits and is known to be associated with degeneration and calcification. Here we analysed samples of herniated disc material and compared it to material taken from non-herniated discs following surgical treatment in dogs. Our clinical approach to these cases allows collection of samples providing a unique opportunity for a case-controlled study such as this, an opportunity which is not available to the human neurosurgeon. We analysed all samples using Fourier transform infrared (FTIR) spectroscopy, as well as a proportion with X-ray diffraction (XRD) and transmission electron microscopy (TEM). FTIR spectra of the majority of herniated samples were consistent with the presence of crystalline hydroxyapatite, whereas most of the non-herniated discs showed spectra consistent with amorphous phosphate material. XRD analysis and TEM confirmed these findings and identified the amorphous material as amorphous calcium phosphate nanoparticle clusters of [~] 20 nm diameter and the crystalline hydroxyapatite material as needles up to 100 nm in length. The differences between the herniated and non-herniated discs indicate that the degenerative process involves a conversion of amorphous calcium phosphate into crystalline hydroxyapatite which precedes and may predispose the disc to herniate.

neuroscience↗