Biology subjects

Kumar, S.

Publications and source records attributed to Kumar, S..

At least 199 records · Page 11Linked to original sources

bioRxiv2025

Assembly, Annotation, and Comparative Analysis of the Mitochondrial Genome of Winged Bean (Psophocarpus tetragonolobus): Insights into Evolutionary Adaptation and Codon Usage Bias

The winged bean (Psophocarpus tetragonolobus) is an underutilized, nutritionally dense legume with significant agricultural potential due to its adaptability to tropical environments and nitrogen-fixing capabilities. This study presents the assembly, annotation, and comparative analysis of its complete mitochondrial genome, spanning 366,925 bp. The genome comprises 64 genes, including 38 protein-coding genes, 12 tRNA genes, and 6 rRNA genes. Codon usage analysis revealed moderate bias in mitochondrial genes, primarily shaped by mutational pressures, while tRNA gene content aligns with codon preferences, optimizing translational efficiency. Phylogenetic analysis placed P. tetragonolobus within the Phaseoleae tribe, clustering closely with Glycine max and Vigna radiata, while synteny analysis highlighted high genomic conservation with related legumes. Ka/Ks ratio analysis underscored strong purifying selection in core mitochondrial genes, reflecting functional constraints essential for energy metabolism. Comparative analyses of mitochondrial and chloroplast genomes revealed distinct evolutionary pressures, with chloroplast genes exhibiting more significant variability influenced by functional roles in photosynthesis. These findings provide valuable insights into the evolution and adaptation of P. tetragonolobus, establishing a foundation for its potential application in sustainable agriculture and crop improvement programs.

genomics↗

bioRxiv2025

Druggable genome CRISPRi screen in 3D hydrogels reveals regulators of cortactin-driven actin remodeling in invading glioblastoma cells

To identify new therapeutic targets that limit glioblastoma (GBM) invasion, we applied druggable-genome CRISPR screens to patient-derived GBM cells in micro-dissectible biomimetic 3D hydrogel platforms that permit separation and independent analysis of core vs. invasive fractions. We identified 12 targets whose suppression limited invasion, of which ACP1 (LMW-PTP) and Aurora Kinase B (AURKB) were validated in neurosphere assays. Proximity labeling analysis identified cortactin as an ACP1- AURKB link, as cortactin undergoes serine phosphorylation by AURKB and tyrosine dephosphorylation by ACP1. Suppression of ACP1 or AURKB in culture and in vivo shifted the balance of cortactin phosphorylation in GBM and reduced actin polymerization and actin-cortactin co-localization. Additional biophysical analysis implicated AURKB in GBM cell adhesion and cortical stiffness, and ACP1 in resistance to mechanical stress and shape plasticity needed for 3D migration. These findings reveal a novel targetable axis that balances kinase and phosphatase activities to regulate actin polymerization during GBM invasion.

cancer biology↗

bioRxiv2025

Oxidative Phosphorylation (OXPHOS) Promotes the Formation and Growth of Melanoma Lung and Brain Metastases

Melanoma mortality is driven by the formation and growth of distant metastases. Here, we interrogated the role of tumor oxidative phosphorylation (OXPHOS) in the formation of distant metastases in melanoma. OXPHOS was the most upregulated metabolic pathway in primary tumors that formed distant metastases in the RCAS-TVA mouse model of spontaneous lung and brain metastases, and in melanoma patients that developed brain or other distant metastases. Knockout of PGC1 in melanocytes in the RCAS-TVA melanoma mouse model had no impact on primary tumor formation, but markedly reduced the incidence of lung and brain metastases. Genetic knockout of a component of electron transport chain complex I, NDUFS4, in B16-F10 and D4M-UV2 murine melanoma cell lines did not impact tumor incidence following subcutaneous, intravenous, or intracranial injection, but decreased tumor burden specifically in the lungs and brain. Together, these data demonstrate that OXPHOS is critical for the formation of metastases in melanoma. STRUCTURED ABSTRACTO_ST_ABSPurposeC_ST_ABSMelanoma mortality is driven by the formation and growth of distant metastases. However, the process and pathogenesis of melanoma metastasis remain poorly understood. Here, we interrogate the role of tumor oxidative phosphorylation (OXPHOS) in the formation of distant metastases in melanoma. Experimental DesignThis study includes (1) new RNA-seq analysis of primary melanomas from patients characterized for distant metastasis events; (2) RNA-seq analysis and functional testing of genetic OXPHOS inhibition (PGC1 KO) the RCAS-TVA model, which is the only existing immunocompetent murine model of autochthonous lung and brain metastasis formation from primary melanoma tumors; and (3) functional experiments of genetic OXPHOS inhibition (NDUFS4 KO) in the B16-F10 and D4M-UV2 murine melanoma cell lines, including evaluation of subcutaneous, lung, and brain metastatic site dependencies. ResultsOXPHOS was the most upregulated metabolic pathway in primary tumors that formed distant metastases in the RCAS-TVA mouse model of spontaneous lung and brain metastases, and in melanoma patients that developed brain or other distant metastases. Knockout of PGC1a in melanocytes in the RCAS-TVA melanoma mouse model had no impact on primary tumor formation, but markedly reduced the incidence of lung and brain metastases. Genetic knockout of a component of electron transport chain complex I, NDUFS4, in B16-F10 and D4M-UV2 murine melanoma cell lines did not impact tumor incidence following subcutaneous, intravenous, or intracranial injection, but decreased tumor burden specifically in the lungs and brain. ConclusionsTogether, these data demonstrate that OXPHOS is critical for the formation of metastases in melanoma. TRANSLATIONAL RELEVANCEMelanoma is the most aggressive form of skin cancer. One hallmark of this disease is a high risk of distant metastasis formation. The process and pathogenesis of metastasis in this disease remain poorly understood and there is controversy regarding the role of oxidative phosphorylation (OXPHOS) in melanoma metastasis. This study incorporates RNAseq analysis of primary melanoma tumors from patients characterized for distant metastasis events, RNAseq analysis of the only existing immunocompetent murine model of autochthonous lung and brain metastasis formation from primary melanoma tumors, and functional testing in multiple syngeneic models of melanoma at different tissue sites. This integrated analysis consistently demonstrates that melanoma OXPHOS promotes distant metastasis to the lungs and brain, two of the most common and clinically relevant sites of melanoma metastasis. This improved understanding of tumor OXPHOS may represent novel vulnerabilities for therapeutics development and surveillance/preventative strategies for melanoma metastasis.

cancer biology↗

bioRxiv2025

Convergent Agonist and Heat Activation of Nociceptor TRPM3

Detecting noxious heat is vital for survival, triggering pain responses that protect against harm1,2. The TRPM3 channel is a key nociceptor for sensing noxious heat and a promising therapeutic target for pain treatment and neurological disorders such as epilepsy3-11. Here, we functionally and structurally characterized TRPM3 in response to diverse stimuli: the synthetic superagonist CIM0216 Ref12, the anticonvulsant antagonist primidone13,14, and heat1,10,15. Our findings reveal that TRPM3 is intrinsically dynamic, with its intracellular domain (ICD) sampling both resting and activated states, though strongly favoring the resting state without stimulation. CIM0216 binds to the S1-S4 domain, inducing conformational changes in the ICD and shifting the equilibrium toward activation. Remarkably, heat induces similar ICD rearrangements, revealing a converged activation mechanism driven by chemical compounds and temperature. This mechanism is supported by functional data showing that mutations facilitating the ICD movement markedly increase the sensitivity of TRPM3 to both chemical and thermal signals. These findings establish a critical role of the ICD in temperature sensing in TRPM3, a mechanism likely conserved across the TRPM family. Finally, we show that primidone binds to the same site as CIM0216 but acts as an antagonist. This study provides a framework for understanding the thermal sensing mechanisms of temperature-sensitive ion channels and offers a structural foundation for developing TRPM3-target therapeutics for pain and neurological disorders.

biophysics↗

bioRxiv2025

Entrenchment and contingency in neutral protein evolution with epistasis

Protein sequence evolution in the presence of epistasis makes many previously acceptable amino acid residues at a site unfavorable over time. This phenomenon of entrenchment has also been observed with neutral substitutions using Potts Hamiltonian models. Here, we show that simulations using these models often evolve non-neutral proteins. We introduce a Neutral-with-Epistasis (NxE) model that incorporates purifying selection to conserve fitness, a requirement of neutral evolution. NxE protein evolution revealed a surprising lack of entrenchment, with site-specific amino-acid preferences remaining remarkably conserved, in biologically realistic time frames despite extensive residue coupling. Moreover, we found that the overdispersion of the molecular clock is caused by rate variation across sites introduced by epistasis in individual lineages, rather than by historical contingency. Therefore, substitutional entrenchment and rate contingency may indicate that adaptive and other non-neutral evolutionary processes were at play during protein evolution.

evolutionary biology↗

bioRxiv2025

Small molecule promoters of endogenous lipid droplet accumulation drive lysophagy

Lipid droplets (LDs) play a central role in regulating metabolism in stress-induced conditions, including one triggered by nutrient deprivation. Unravelling the protein networks involved in the biogenesis of LDs and their causative and functional roles in health and disease continue to evolve. To this cause, genetic manipulation of the lipid metabolic network or supplementation of high fat diet/ oleic acid (OA) are the traditional routes for voluntarily triggering LDs formation in cells and animals. We developed a screening platform for the identification of new LDs inducers, and our primary screening of various fatty acids identified linoleic acid (LOA, DUFA) as a better tool than OA (MUFA) in promoting LDs formation. The screening and validation discovered new small molecule-based tools for promoting a rapid organization of endogenous lipids into droplets in multiple cell types. Notably, our mass spectral lipidomics analysis presented the overproduction of phosphatidylcholines and small triglycerides, a hallmark of LDs. Mechanistic investigations of our lead molecules highlighted lipid peroxidation and ATP depletion through mitochondrial impairment in cells, which could serve as chemical cues for driving the fusion of cellular lipids into LDs. Finally, we uncovered the abrupt levels of LDs formation induced by our molecules promoted lysophagy in cancer cells to prevent their proliferation. Collectively, our work introduces new small molecules as powerful tools for reliably promoting LDs accumulation for studying their roles in biology, and we demonstrate the over accumulation of LDs prevent cancer cell proliferation, movement, and colonization.

cancer biology↗

bioRxiv2025

A V-ATPase complex, required for biogenesis of a lysosome-related organelle (LRO) in Tetrahymena thermophila, disassembles during LRO maturation

Lysosome-related organelles (LROs) are a heterogeneous family of organelles found in many cell types, whose significant similarities to lysosomes include their acidification by vacuolar-type ATPases (V-ATPases). However, some organelles with hallmarks of LROs are nonetheless non-acidic. Here we investigate this poorly understood phenomenon using the ciliate Tetrahymena thermophila, which have prominent secretory LROs called mucocysts. Using three different approaches, we show that mature mucocysts, poised for exocytosis, are non-acidic. Nonetheless, a specific V-ATPase a-subunit is targeted to mucocysts and is indispensable for mucocyst biogenesis and secretion. Consistent with this, immature mucocysts in the cytoplasm appear to be transiently acidified. The stage specificity of acidification can be explained by our finding that several canonical V-ATPase subunits are present in the immature, but not mature, mucocysts. Our data argue that a mucocyst-specific V-ATPase complex is targeted to newly-forming mucocysts and subsequently disassembles at a late stage in Tetrahymena LRO maturation, and that this depends on the AP-3 trafficking adaptor.

cell biology↗

bioRxiv2025

MiRNA-501-3p and MiRNA-502-3p: A Promising Biomarker Panel for Alzheimer's Disease

INTRODUCTIONAlzheimers disease (AD) lacks a less invasive and early detectable biomarker. Here, we investigated the biomarker potential of miR-501-3p and miR-502-3p using different AD sources. METHODSMiR-501-3p and miR-502-3p expressions were evaluated in AD CSF exosomes, serum exosomes, familial and sporadic AD fibroblasts and B-lymphocytes by qRT-PCR analysis. Further, miR-501-3p and miR-502-3p expressions were analyzed in APP, Tau cells and media exosomes. RESULTSMiR-501-3p and miR-502-3p expressions were significantly upregulated in AD CSF exosomes relative to controls. MiRNA levels were high in accordance with amyloid plaque and NFT density in multiple brain regions. Similarly, both miRNAs were elevated in AD and MCI serum exosomes compared to controls. MiR-502-3p expression was high in fAD and sAD B-lymphocytes. Finally, miR-501-3p and miR-502-3p expression were elevated intracellularly and secreted extracellularly in response to APP and Tau pathology. DISCUSSIONThese results suggest that miR-501-3p and miR-502-3p could be promising biomarkers for AD.

neuroscience↗

bioRxiv2025

Evolutionary sparse learning with paired species contrast reveals the shared genetic basis of convergent traits

Cases abound in which nearly identical traits have appeared in distant species facing similar environments. These unmistakable examples of adaptive evolution offer opportunities to gain insight into their genetic origins and mechanisms through comparative analyses. Here, we present a novel comparative genomics approach to build genetic models that underlie the independent origins of convergent traits using evolutionary sparse learning. We test the hypothesis that common genes and sites are involved in the convergent evolution of two key traits: C4 photosynthesis in grasses and echolocation in mammals. Genetic models were highly predictive of independent cases of convergent evolution of C4 photosynthesis. These results support the involvement of sequence substitutions in many common genetic loci in the evolution of convergent traits studied. Genes contributing to genetic models for echolocation were highly enriched for functional categories related to hearing, sound perception, and deafness (P < 10-6); a pattern that has eluded previous efforts applying standard molecular evolutionary approaches. We conclude that phylogeny-informed machine learning naturally excludes apparent molecular convergences due to shared species history, enhances the signal-to-noise ratio for detecting molecular convergence, and empowers the discovery of common genetic bases of trait convergences.

evolutionary biology↗

bioRxiv2025

Hypoxia-Mediated Molecular Interactions of Tissue-Specific Mesenchymal Stem Cells Drive Metabolic Reprogramming and Immunomodulation in Acute Graft-versus-Host Disease

BackgroundMesenchymal stem cells (MSCs) mediate immunomodulation through various mechanisms, including apoptosis, efferocytosis, and mitochondrial transfer. Our study investigates the impact of hypoxia preconditioning on the immune metabolic reprogramming and immunomodulatory potential of MSCs in acute graft-versus-host disease (aGVHD). Additionally, we explored the differential immunomodulatory effects of tissue-specific MSCs, specifically bone marrow (BM) and Whartons Jelly (WJ), and elucidated the mechanisms underlying variability in their therapeutic efficacy. MethodsMSCs were isolated from BM and WJ and subjected to hypoxia preconditioning. Their immunometabolic programming potential was assessed by evaluating T-cell proliferation, regulatory T-cell (Treg) induction, effector T-cell differentiation toward Th2, Th9 phenotypes, and macrophage polarization, T-cell bioenergetics in the direct co-culture systems. ResultsWJ-MSCsHYP exhibited superior immunomodulatory properties compared to BM-MSCsHYP, by inhibiting T-cell proliferation, enhancing Treg induction, and promoting anti-inflammatory macrophage polarization. WJ-MSCsHYP demonstrated enhanced mitochondrial transfer to T-cell, improving mitochondrial health, reducing ROS, and promoting oxidative phosphorylation, leading to immune homeostasis. Unlike BM-MSCs, WJ-MSCs exhibited higher rates of apoptosis, which facilitated immune modulation through mechanisms independent of efferocytosis. ConclusionOur findings highlight that WJ-MSCsHYP is a superior candidate for aGVHD by utilizing apoptosis, mitochondrial transfer, and metabolic reprogramming to achieve immune regulation.

immunology↗

bioRxiv2025

Identification and Characterization of Novel Outer Membrane Proteins of Brachyspira pilosicoli

Brachyspira pilosicoli is a pathogenic, Gram-negative, spirochete bacterium that causes intestinal spirochetosis (IS) in birds, pigs, and humans and is distributed worldwide. This anaerobic intestinal bacterium colonizes the large intestine, potentially leading to colitis, diarrhoea, and decreased growth rate. Outer membrane proteins of Gram-negative bacteria play crucial roles in adhesion and host-pathogen interaction, helping the bacteria to evade the host immune system, and enhancing their virulence. However, B. pilosicoli outer membrane proteins are yet to be identified and characterized. Here, we report the computational discovery of 42 outer membrane {beta}-barrel (OMBB) proteins in B. pilosicoli proteome predicted using a consensus-based computational framework. {beta}-barrel architectures of the predicted proteins were confirmed by generating AlphaFold 3-based structural models. Structure-based functional annotation predicted putative functions for the identified OMBB proteins, including BamA homolog involved in folding and membrane insertion of OMPs, LptD homolog involved in transport of lipopolysaccharides into the OM, efflux pumps, transporters, enzymes, diffusion channels, and porins. Sequence variations across nine strains of B. pilosicoli were identified and mapped onto structural models, revealing that many of the variations were present on the surface exposed loop regions of the {beta}-barrel structures. Our in-silico analysis has identified 42 OMBB proteins, including homologs of BamA, LptD, TolC, TonB-dependent receptors, CsgG. Seven of these were identified as hypothetical proteins. Computational characterization of the predicted OMBB proteins offer insights into their potential roles in physiology, virulence, and disease pathogenesis, highlighting their potential for applications in diagnostics, vaccine development, or therapeutic interventions.

bioinformatics↗

bioRxiv2024

Exploring the microbiome of black soil from Central India and the impacts of agricultural practices on soil microbial communities

BackgroundHealthy microbial communities in soil are indispensable for agricultural productivity. The diversity in physiochemical properties, their geological origins, and local climatic conditions lead to the evolution of diverse soil types with distinct bacterial communities. Despite a global distribution, the microbial composition of black soil remains largely unexplored. Here, we investigate the composition, diversity, and functional potential of bacterial communities in black soils from central India. We also examine how common agricultural disturbances like fertilizer application and the growth of crops impact soil communities in a three-year longitudinal field experiment involving three crops. ResultsThe black soil core microbiome was dominated by phyla Actinobacteriota and Proteobacteria members. Analysis of bacterial composition in vermicompost revealed the presence of plant growth-promoting taxa. Among the assessed cropping practices, crop growth significantly affected microbiome composition in soil. We also observed the impact of crops in selectively enhancing the microbiome in the soil, highlighting the role of plant species in shaping community structure. During this experiment, fertilizer treatment did not influence the microbiome significantly. Overall, the data showed the vulnerability and resilience of soil bacterial communities to agricultural disturbances. ConclusionThe microbial composition of black soil in central India was revealed, and the effects of agricultural management on black soil microbial composition were evaluated. The prominent impact of individual plants showed the role of plant-root microbiome interactions in shaping the composition of soil communities. Fertilizer application did not show significant effects during the field experiment, reflecting the resilience of soil communities to short-term disturbances. We also shed light on the functional metagenomic potential of black soil communities and the effect of common agricultural disturbances in a prominent soil type.

microbiology↗

bioRxiv2024

A Blueprint of Sex-Specific Neuronal Regulation in the C. elegans Nervous System at Single-Cell Resolution

Sex-specific behaviors are often attributed to differences in neuronal wiring and molecular composition, yet how genetic sex shapes the molecular architecture of the nervous system at the individual neuron level remains unclear. Here, we use single-cell RNA sequencing to profile the transcriptome of sex-shared neurons in adult Caenorhabditis elegans males and hermaphrodites. We uncover widespread molecular dimorphism across the nervous system, including in previously unrecognized neuron-types such as the touch receptors. Neuropeptides and signaling-related genes exhibit strong sex-biased expression, particularly in males, reinforcing the notion that neuropeptides are crucial for diversifying connectome outputs. Despite these differences, neurotransmitter identities remain largely conserved, indicating that functional dimorphism arises through modulatory, not identity-defining, changes. We show that sex-biased expression of neurotransmitter-related genes correlates with bias in outgoing synaptic connectivity and identify regulatory candidates for synaptic wiring, including both shared and sex-specific genes. This dataset provides a molecular framework for understanding how subtle regulatory differences tune conserved circuits to drive sex-specific behaviors.

neuroscience↗

bioRxiv2024

Proteomic Characterization and Molecular Mechanism of Goat Whey Protein-Derived Bioactive Peptides as Pancreatic Lipase and alpha-Amylase Inhibitors

Goat whey protein (GWP) is recognized as a valuable source of bioactive peptides with significant health-promoting properties. In this study, GWP was enzymatically hydrolyzed using a combination of gastrointestinal enzymes-- pepsin, trypsin, and chymotrypsin--to generate peptides. These peptides were identified through high-resolution liquid chromatography-mass spectrometry (HR-LC/MS), resulting in library of 2,883 peptides with lengths ranging from 6 to 44 amino acids. Among them, 40 peptides were predicted to exhibit high bioactivity scores (0.90-1) based on PeptideRanker analysis, with 28 of these being classified as nontoxic. Molecular docking simulations were employed to investigate the interactions of these peptides with pancreatic lipase and -amylase to screening of inhibitors, these two enzymes critical in lipid and carbohydrate metabolism. Several peptides demonstrated strong binding affinities, suggesting their potential as enzyme inhibitors. Notably, peptides WPGIMR and WQDGSWQF showed the highest binding affinity for pancreatic lipase, while AAPFIWL and WQDGSWQF exhibited significant interactions with -amylase. These results shed light on the molecular mechanisms underlying the inhibitory activities of whey protein-derived peptides. They highlight their potential applications as functional food ingredients or natural therapeutic agents for managing metabolic disorders such as obesity and diabetes, advancing the understanding of whey protein hydrolysates in modulating key metabolic enzymes.

bioinformatics↗

bioRxiv2024

Conserved heavy/light contacts and germline preferences revealed by a large-scale analysis of natively paired human antibody sequences and structural data.

Antibody next-generation sequencing (NGS) datasets have become crucial to develop computational models addressing this successful class of therapeutics. Although antibodies are composed of both heavy and light chains, most NGS sequencing depositions provide them in unpaired form, reducing their utility. Here we introduce PairedAbNGS, a novel database with paired heavy/light antibody chains. To the best of our knowledge, this is the largest resource for paired natural antibody sequences with 58 bioprojects and over 14 million assembled productive sequences. We make the database accessible at http://naturalantibody.com/paired-ngs as a valuable tool for biological and machine-learning applications. Using this dataset, we investigated heavy and light chain variable (V) gene pairing preferences and found significant biases beyond gene usage frequencies, possibly due to receptor editing favoring less autoreactive combinations. Analyzing the available antibody structures from the Protein Data Bank, we studied conserved contact residues between heavy and light chains, particularly interactions between the CDR3 region of one chain and the FWR2 region of the opposite chain. Examination of amino acid pairs at key contact sites revealed significant deviations of amino acids distributions compared to random pairings, in the heavy chains CDR3 region contacting the opposite chain, indicating specific interactions might be crucial for proper chain pairing. This observation is further reinforced by preferential IGHV-IGLJ and IGLV-IGHJ pairing preferences. We hope that both our resources and the findings would contribute to improving the engineering of biological drugs.

immunology↗

bioRxiv2024

CD38 biallelic loss is a recurrent mechanism of resistance to anti-CD38 antibodies in multiple myeloma.

Monoclonal antibodies targeting CD38 are a therapeutic mainstay in multiple myeloma (MM). While they have contributed to improved outcomes, most patients still experience disease relapse, and little is known about tumor-intrinsic mechanisms of resistance to these drugs. Antigen escape has been implicated as a mechanism of tumor cell evasion in immunotherapy. Yet, it is unknown whether MM cells can develop permanent resistance to anti-CD38 antibodies by acquiring genomic events leading to biallelic disruption of the CD38 gene locus. Here, by using whole genome and whole exome sequencing data from 701 newly diagnosed patients, 67 patients at relapse with naivety to anti-CD38 antibodies, and 50 patients collected at relapse following anti-CD38 antibodies. We report a loss of CD38 in 20% (10/50) of patients post-CD38 therapy, three of which exhibited a loss of both copies. Two of these cases showed convergent evolution where distinct subclones independently acquired similar advantageous variants. Functional studies on missense mutations involved in biallelic CD38 events revealed that two variants, L153H and C275Y, decreased binding affinity and antibody-dependent cellular cytotoxicity of the commercial antibodies Daratumumab and Isatuximab. However, a third mutation, R140G, conferred selective resistance to Daratumumab, while retaining sensitivity to Isatuximab. Clinically, patients with MM are often rechallenged with CD38 antibodies following disease progression and these data support a role for next generation sequencing to guide treatment selection.

cancer biology↗

bioRxiv2024

Multimodal normative modeling in Alzheimer Disease with introspective variational autoencoders

Normative models in neuroimaging learn patterns of healthy brain distributions to identify deviations in disease subjects, such as those with Alzheimers Disease (AD). This study addresses two key limitations of variational autoencoder (VAE)-based normative models: (1) VAEs often struggle to accurately model healthy control distributions, resulting in high reconstruction errors and false positives, and (2) traditional multimodal aggregation methods, like Product-of-Experts (PoE) and Mixture-of-Experts (MoE), can produce uninformative latent representations. To overcome these challenges, we developed a multimodal introspective VAE that enhances normative modeling by achieving more precise representations of healthy anatomy in both the latent space and reconstructions. Additionally, we implemented a Mixture-of-Product-of-Experts (MOPOE) approach, leveraging the strengths of PoE and MoE to efficiently aggregate multimodal information and improve abnormality detection in the latent space. Using multimodal neuroimaging biomarkers from the Alzheimers Disease Neuroimaging Initiative (ADNI) dataset, our proposed multimodal introspective VAE demonstrated superior reconstruction of healthy controls and outperformed baseline methods in detecting outliers. Deviations calculated in the aggregated latent space effectively integrated complementary information from multiple modalities, leading to higher likelihood ratios. The model exhibited strong performance in Out-of-Distribution (OOD) detection, achieving clear separation between control and disease cohorts. Additionally, Z-score deviations in specific latent dimensions were mapped to feature-space abnormalities, enabling interpretable identification of brain regions associated with AD pathology.

neuroscience↗

bioRxiv2024

MEGA12: Molecular Evolutionary Genetic Analysis version 12 for adaptive and green computing

We introduce the 12th version of the Molecular Evolutionary Genetics Analysis (MEGA) software. This latest version brings many significant improvements by reducing the computational time needed for selecting optimal substitution models and conducting bootstrap tests on phylogenies using maximum likelihood (ML) methods. These improvements are achieved by implementing heuristics that minimize likely unnecessary computations. Analyses of empirical and simulated datasets show substantial time savings by using these heuristics without compromising the accuracy of results. MEGA12 also implements an evolutionary sparse learning approach to identify fragile clades and associated sequences in evolutionary trees inferred through phylogenomic analyses. In addition, this version includes fine-grained parallelization for ML analyses, support for high-resolution monitors, and an enhanced Tree Explorer. The MEGA12 beta version can be downloaded from https://www.megasoftware.net/beta_download.

evolutionary biology↗