Biology subjects

Rahman, T.

Publications and source records attributed to Rahman, T..

At least 19 recordsLinked to original sources

bioRxiv2026

Charged residues, not aromatics, are the universal topological hubs of intrinsically disordered proteins

The stickers and spacers model assigns aromatic residues as the primary drivers of IDP phase separation predicting that aromatics also occupy topologically central hub positions in IDP contact networks. Analysing a proteome wide simulation dataset spanning six taxa we find that charged residues not aromatics are the universal network hubs with glutamate the strongest hub in every taxon and aromatics predominantly anti hub. All IDP contact networks are universally positively assortative a sequence encoded topological law in which hub identity is positionally encoded. That hub character requires charge carrier geometry rather than charge alone is demonstrated by a natural experiment in which arginine and lysine carry identical net charge yet arginine is a universal hub while lysine is a near neutral spacer (Cohens d = 2.456) confirmed by force field cross validation and wet lab condensate miscibility data. Charge matched synuclein K[->]R slab simulations reveal that hub character governs condensate density and thermodynamic stability without altering the nucleation threshold connecting single chain topology to multi chain phase thermodynamics. Tyrosine hub character tracks the prokaryote eukaryote boundary with evolutionary encoding the only explanation to survive direct testing. Together these results establish hub identity and sticker identity as complementary organisational principles of disordered sequences governed by distinct residue chemistries and shaped independently by selection.

biophysics↗

bioRxiv2026

BENDER: A Cross-taxon IDP Simulation Database Reveals Conserved Sequence-Ensemble Laws Across the Tree of Life

Intrinsically disordered proteins and regions are found across all kingdoms of life, yet the computational characterisation of their conformational ensembles has remained almost entirely confined to the human proteome. Whether the force fields used to generate them remain accurate for taxonomically distant organisms, and whether the sequence ensemble relationships they reveal transfer across taxa well enough to improve prediction on phylogenetically held-out organisms, are open questions. Here we introduce BENDER, a dataset of 11,533 IDP sequences spanning 13 taxonomic groups, each simulated under CALVADOS 2 molecular dynamics and annotated with ensemble-level geometric and novel contact-network properties, together with per-sequence pi pi and cation pi contact frequencies linked to phase-separation propensity. We show that CALVADOS 2 ensembles agree strongly with an orthogonal structural reference across the full dataset,with both held out taxa performing above the dataset median, and that direct comparison against a second independently parameterized force field reveals no systematic scaling-exponent bias. We find that cross taxon training data improves out-of-distribution ensemble prediction in two independent architectures despite training on one third the data. Positive degree assortativity is conserved across all taxonomic groups, suggesting that hub topology in disordered protein contact networks is a conserved physical feature of sequence-encoded disorder rather than an evolutionary contingency

biophysics↗

bioRxiv2026

Genomic Surveillance of Respiratory Syncytial Virus among Patients with Acute Respiratory Infection through Hospital-Based Influenza Surveillance Platforms in Bangladesh, August 2024-December 2025

BackgroundRSV is one of the major contributing factors of acute lower respiratory tract infection among younger children who are below five years old; over 95% of the burden falls on low- and middle-income countries. Bangladesh faces a substantial pediatric RSV burden yet lacks post-pandemic whole-genome data. Approval of nirsevimab and maternal vaccines (Arexvy, Abrysvo), underscores the need for region-specific genomic surveillance to inform prevention and control strategies. ObjectivesTo conduct the whole-genome sequencing (WGS) study of RSV in Bangladeshi population, characterizing viral genotypic diversity, phylogenetics, F and G protein mutations, glycosylation dynamics, while generating and depositing complete genomes in GISAID. MethodsBetween August 2024 and December 2025, of 1,390 RSV-positive specimens (Ct [≤]35), 59 high-viral-load samples (Ct [≤]25) were selected for whole-genome sequencing using Oxford Nanopore Technology (ARTIC primers), assembled with MIRA v2.0.0, and analyzed via Augur/Nextstrain, Nextclade, and NetNGlyc/NetOGlyc 4.0. ResultsOf 11,874 patients, 1,390 (11.7%) were RSV-positive (RSV-A 94.6%). From selected 59 RSV-positive cases, 49 high-quality genomes were generated (83.1% pass; 94.7% completeness; median depth 1,500x, range 443-4,948): 43 RSV-A (ON1; A.D.3.7 63%, A.D.3 19%, A.D.3.12 9%, A.D.3.1 7%, A.D.1.11 2%) and 6 RSV-B (BA9/B.D.E.1). S276N at antigenic site II (34.9% RSV-A) and S389P in all RSV-B were detected; neither confers resistance to nirsevimab or palivizumab. An F protein N75 N-glycosylation site was fixed in all RSV-A; RSV-B acquired HVR2 N256 glycan in 67%. All 49 genomes were deposited in GISAID. ConclusionThis WGS study of RSV in the Bangladeshi population confirms LMIC nanopore surveillance feasibility, multi-lineage co-circulation, and intact conservation of all vaccine and antibody targets. Progressive glycan remodeling warrants monitoring. These findings establish a genomic baseline to guide nirsevimab and maternal vaccine deployment in Bangladesh, with direct relevance for RSV surveillance programs across South Asia.

evolutionary biology↗

bioRxiv2026

Do AI Structure Predictors Capture Bound-State Disorder? A Benchmark on Fuzzy Protein Complexes

Fuzzy protein complexes, in which an intrinsically disordered protein (IDP) retains conformational disorder upon binding, pose a fundamental challenge for structure predictors trained on ordered systems, where crystal structures capture only the most ordered ensemble snapshot, making standard benchmarking metrics misleading. Here, we present the first systematic evaluation of AlphaFold3 (AF3), AlphaFold2-Multimer (AF2MM), Chai-1, and Boltz-2 on a curated dataset of fuzzy complexes from FuzDB, benchmarked against DockQ against PDB structures and NOE violation rates against manually curated BMRB restraint files, the first comprehensive collection of this kind. Across all four predictors, approximately 30% of NOE restraints were violated with nearly identical distributions regardless of predictor architecture or training data. DockQ scores fell uniformly within the Acceptable range, with AF3 marginally higher but exhibiting NOE violation rates equivalent to the weakest-performing model. Ensemble-level analysis using a first-principles implementation of the Hadzi thermodynamic model revealed that AF3 uniquely achieves near-zero mean helicity bias, in contrast to systematic overconfidence in the other predictors, yet all four models show poor per-residue helicity correlation with thermodynamic expectations. DockQ rankings reflect training data similarity to crystal structures rather than physical accuracy, and no current predictor captures fuzzy complex ensemble behavior. The FuzzyBench-NOE dataset, comprising NOE restraint files, predicted structures, interface hotspot annotations, and Hadzi-DSSP analysis outputs, is released on Zenodo (https://doi.org/10.5281/zenodo.20470556). Significance StatementNo benchmark exists for fuzzy protein complexes, where IDPs retain disorder upon binding. We show that four state-of-the-art structure predictors violate 30% of experimental NMR distance restraints invariantly regardless of architecture, while DockQ, the standard metric, is entirely uncorrelated with this failure. Ensemble-level analysis using the Ha[d]zi thermodynamic model reveals systematic helicity overconfidence across all predictors. Taken together, our findings imply that standard geometric metrics are fundamentally misleading for disordered systems, thus necessitating ensemble-aware evaluation.

biophysics↗

bioRxiv2026

A unified ensemble-allosteric framework reconciles gain- and loss-of-function disease mutations in the IP3 receptor

Missense mutations in multidomain signalling proteins often produce divergent functional phenotypes despite preserved structural integrity, posing a fundamental challenge for interpreting human genetic variation. This problem is exemplified by the type 1 inositol 1,4,5-trisphosphate receptor, IP3R1, the principal neuronal IP3-gated Ca2+ release channel and a recurrent locus of pathogenic variation in ITPR1-associated ataxias and neurodevelopmental disorders. Although disease mutations cluster within the N-terminal suppression domain (SD) and IP3-binding core (IBC), how neighbouring substitutions can drive either gain- or loss-of-function remains unresolved. Here, we establish a unified ensemble-allosteric framework for IP3R1 channelopathy by integrating mutational constraint analysis, ensemble structural modelling, adaptive molecular dynamics, Markov state modelling and dynamical network analysis. We show that pathogenic variants preferentially occupy a stability-preserving regime, perturbing local sequence-structure compatibility without inducing global structural collapse. Thus, disease arises not from loss of fold, but from corruption of the conformational and allosteric logic that links ligand recognition to channel gating. The IP3-binding pocket forms a discrete multistate ensemble whose populations, physicochemical properties and kinetic connectivity are selectively remodelled by mutation. The loss-of-function variant R269W disrupts a cationic IP3-coordinating site, enriches non-permissive binding-pocket states and diverts conformational exchange through indirect kinetic routes. By contrast, the gain-of-function variant R36C preserves local pocket competence but weakens suppressor-domain restraint by rerouting long-range communication through extended, less efficient allosteric pathways. These findings reconcile opposing disease phenotypes within a single mechanistic model, showing that pathogenic variation can be encoded in altered ensemble probabilities and information flow rather than in static structural lesions.

biophysics↗

bioRxiv2026

A Single-Cell Peripheral Immune Atlas Spanning High-Risk Lesions to Invasive Breast Cancer in Black and White Women

Black women are at risk for breast cancer nearly a decade before women of other racial groups for unclear reasons. Because immune responses influence cancer initiation and progression, we performed single-cell RNA sequencing of peripheral blood mononuclear cells from non-Hispanic Black (NHB) and non-Hispanic White (NHW) women with high-risk breast lesions, ductal carcinoma in-situ, and invasive breast cancer. Race-associated transcriptional differences were observed across all disease states and were most pronounced in invasive disease. Computational analyses, supported by flow cytometric protein analysis, revealed enrichment of chronic inflammation, immune regulatory programs, and immune aging pathways in NHB cancer patients, particularly in monocytes, dendritic cells, CD4+ T cells, and B cells. From these data, we derived a population-level immune signature (IMM-POP) comprising genes differentially enriched in this subset of immune cells from NHB breast cancer patients. IMM-POP correlates with an immunosuppressive signature in external breast cancer datasets. We thus provide a single-cell peripheral immune atlas integrating race and breast disease state. SignificanceThis study revealed race-specific peripheral immunity features in precancerous and invasive breast cancers: Black patients exhibited features of chronic inflammation and immune aging compared with White patients, suggesting immune weathering and providing insights for studying early onset of breast cancer in Black patients.

cancer biology↗

bioRxiv2026

Therapeutic knockdown of MLKL reduces diet-induced obesity and improves insulin signalling in mature adipocytes

Obesity affects one in three adults and is complicated by adipose inflammation, lipotoxicity and cell death. We previously identified RIPK1 as a genetic determinant of human obesity risk and adipose inflammation. Because RIPK1 is the apical kinase in the necroptosis pathway upstream of RIPK3 and the executioner protein MLKL, and emerging evidence links MLKL to lipid metabolism, MLKL has surfaced as a potential metabolic regulator. However, conflicting findings in Mlkl knockout mice fed a high fat diet have left its therapeutic relevance unresolved. MLKL has not been previously targeted through therapeutic knockdown in vivo in the context of diet-induced obesity. Here, we evaluated two independent MLKL antisense oligonucleotides (ASOs) in high fat diet (HFD)-fed C57BL/6J mice. In a 24-week progression model, MLKL ASO markedly reduced body weight, fat mass and hepatic steatosis compared with controls, while preserving lean mass. MLKL knockdown also lowered the respiratory exchange ratio, indicating a shift toward increased fat oxidation. In the intervention model, once obesity was established after 12 weeks of HFD feeding, both MLKL ASOs, and similarly, two independent RIPK1 ASOs, reversed weight gain and improved systemic glucose control. In vitro, MLKL-CRISPR/Cas9 knockout blocked 3T3-L1 adipogenesis, indicating a requirement for MLKL during adipocyte differentiation. However, in mature adipocytes, MLKL siRNA reduced palmitic acid-induced lipid accumulation, increased isoprenaline-stimulated lipolysis, and prevented TNF-mediated suppression of insulin-mediated AKT signalling and glucose uptake. Collectively, these findings demonstrate that partial MLKL suppression reprograms whole-body energy metabolism, enhances insulin sensitivity and limits diet-induced adiposity. MLKL, therefore, represents a promising and mechanistically novel therapeutic target for obesity and insulin resistance.

physiology↗

bioRxiv2026

Evolutionary Remodeling of Ubiquinone Biosynthesis in Toxoplasma gondii Reveals an Essential Bi-functional Monooxygenase

Apicomplexan parasites like Toxoplasma gondii harbor a highly divergent mitochondrial proteome, much of which remains uncharacterized despite its essentiality for parasite survival. One such critical pathway is ubiquinone (UQ) biosynthesis. Here, we characterize the UQ synthesis machinery in T. gondii and show that conserved enzymes, TgCoq3 and TgCoq5, are essential for growth and mitochondrial function, forming a multi-protein complex. Using proximity labeling and subcellular fractionation, a strategy suited for detecting proteins of low abundance, we identify TgCoqFAD, a unique FAD-dependent monooxygenase required for UQ synthesis. Unlike canonical eukaryotic systems that employ multiple monooxygenases to modify specific carbons on the UQ aromatic ring, TgCoqFAD catalyzes two distinct hydroxylation steps, an activity not previously reported in eukaryotes. Molecular docking and chemical screening identified TgCoqFAD inhibitors that impair tachyzoite growth and bradyzoite viability. These findings reveal a streamlined and divergent UQ biosynthesis pathway in apicomplexans and establish TgCoqFAD as a promising antiparasitic target.

biochemistry↗

bioRxiv2025

AlphaFold3 and Intrinsically Disordered Proteins: Reliable Monomer Prediction, Unpredictable Multimer Performance

AlphaFold3 represents a major advance in protein structure prediction, yet its performance on intrinsically disordered proteins remains uncharacterized. We present the first systematic evaluation of AF3 on disordered systems, revealing a striking dichotomy. For monomers, AF3s pLDDT scores reliably predict disorder (MCC: 0.693), matching AlphaFold2 and rivaling dedicated predictors. This consistency across fundamentally different architectures confirms that disorder prediction emerges from training data, not model design. For multimers, the picture grows complex. Despite comparable aggregate performance (mean DockQ: 0.563 vs 0.571), AF3 and AF2 achieve these results through fundamentally different mechanisms. Conventional structural features explain 58% of AF2s variance but only 42% of AF3s. Users cannot predict when AF3 will succeed or fail from interface properties alone. On disorder-to-order transitions (MFIB benchmark), both models perform equally well, successfully predicting final folded states. Yet seed variance analysis reveals AF3s failures are deterministic: the model converges to identical structures across independent runs, whether correct or incorrect, indicating rigid structural priors override available information. Our findings establish AF3 as reliable for the prediction of monomer disorder but unpredictable for multimers. Architectural innovation alone cannot overcome training data bias. Progress demands disorder-enriched datasets and ensemble sampling, not merely novel architectures.

bioinformatics↗

bioRxiv2025

A GCN1-independent activator of the kinase GCN2

Mutations of EIF2AK4, which encodes the eIF2 kinase GCN2, cause a severe inherited form of pulmonary hypertension called pulmonary veno-occlusive disease (PVOD). Some pathogenic variants of GCN2 are amenable to pharmacological reactivation by low concentrations of ATP-pocket binding inhibitors. Kinase inhibition at modestly elevated concentrations limits the clinical utility of these drugs against PVOD. We therefore performed an in cellulo chemical screen for GCN2 activators and identified three structurally distinct compounds with low micromolar stimulatory activities. Unlike previously described GCN2 activators, one of these molecules activated GCN2 independently of GCN1. Modelling supported by structure activity screens suggested it binds within the ATP-pocket of GCN2, but unlike existing ligands does not protrude inward into the allosteric pocket or outward into the solvent. This overcomes a key requirement of other GCN2 activators.

cell biology↗

bioRxiv2025

Discovering new amyloid-like peptides using all-atom simulations and artificial intelligence.

Establishing the fundamental relationships between peptide sequences and fibril formation is critical both for understanding protein misfolding processes and for guiding biomaterial design. Here, we combine all-atom molecular dynamics (MD) simulations with artificial intelligence (AI) to investigate how subtle variations in the arrangement of a short peptide sequence affect its propensity to form fibrils. Our results show that small shifts in the distribution of hydrophobic residues and charge clusters can significantly influence both the nucleation rate and the stability of cross-{beta} structures. To rapidly extend this analysis over a wide sequence space, we developed an active learning-enhanced framework--Machine Learning for Molecular Dynamics (ML4MD)--that iteratively refines its predictions based on MD-derived aggregation data. ML4MD efficiently screens numerous peptide permutations and guides the discovery of previously unrecognized fibril-prone sequences, achieving an area under the receiver operating characteristic (ROC) curve (AUC) of 0.939. Overall, ML4MD streamlines the rational design of amyloid-like peptides by integrating detailed atomistic simulations with rapid and high-accuracy ML predictions.

biophysics↗

bioRxiv2024

Functional characterization of the store-operated calcium entry pathway in naked mole-rat cells

Naked mole-rats (NMRs, Heterocephalus glaber) are highly unusual rodents exhibiting remarkable adaptations to their subterranean habitat and resistance to developing various age-related diseases such as those related to abnormal cell proliferation or cancer, neurodegeneration and inflammation. In other rodents, as well as humans, a ubiquitous Ca2+ influx pathway, namely the store-operated Ca2+ entry (SOCE), has been implicated in all these diseases. SOCE is triggered by intracellular Ca2+ store depletion resulting in interaction of Stim proteins with Orai proteins, the putative homologs of which appear be present in the NMR genome, but no characterisation of SOCE in NMRs has yet been conducted. In this study, we provide the first functional and pharmacological characterization of SOCE in NMR cells using both excitable and non-excitable cells.

cell biology↗

bioRxiv2024

Phytochemical and Therapeutic Evaluation of Ficus nervosa Leaves: Impact on Oxidative Stress, Analgesic, Diarrhea, and Blood Clotting in a Mice Model

BackgroundFicus nervosa (FN) leaves have been traditionally used in ethnomedicine for the treatment of various ailments, this study aimed to investigate the cold methanol extract of FN for its antioxidant, analgesic, antidiarrheal, and thrombolytic properties, along with preliminary phytochemical screening to identify key bioactive compounds and acute toxicity test to assess the safety profile of the extract. MethodsIn this study, we conducted an initial investigation to identify the major phytochemical groups present in the leaves of Ficus nervosa. Using conventional phytochemical screening methods on cold methanol extracts, we consistently identified phenols and flavonoids as the predominant bioactive compounds. Following this phytochemical characterization, we assessed the biological activities of the extracts across multiple models. The antioxidant capacity and free radical-scavenging ability of the leaves extracts were evaluated using the DPPH (2,2-Diphenyl-1-picrylhydrazyl) assay. Analgesic efficacy was assessed through acetic acid-induced writhing and electrically induced heat nociception tests. Antidiarrheal effects were examined using castor oil-induced diarrhea and an enteropooling assay in mice. To determine thrombolytic activity, we conducted a human blood clot lysis assay, while preclinical toxicity testing was conducted to evaluate the safety profile of the extracts. Statistical analysis was performed using one-way ANOVA with Duncans Multiple Range Test (DMRT) to validate the significance of the results. ResultsThe study of Antioxidant Potential, the ESFs (ethyl soluble fraction) antioxidant capacity, measured at 409.91 {+/-} 0.59 mg of ascorbic acid equivalent (AAE)/g in the phosphomolybdenum assay, signifies a strong free radical scavenging activity. This level of antioxidant action, alongside a potent free radical neutralizing concentration (IC of 9.99 {+/-} 0.41 g/mL), suggests that the ESF could effectively counteract oxidative stress. Analgesic activity at 500 mg/kg, FN extract produced a 68.05% suppression of writhing in mice, which was statistically significant (p<0.001) compared to standard morphines (74.05%) inhibition. The extract also significantly prolonged reaction latency to thermal-induced pain in hotplate model. The antidiarrheal activity showed 52.09% and 59.0% suppression of defecation, compared to standard loperamides 73.1%, with statistically significant p-values (p<0.04 and p<0.01). These effects, combined with the notable decrease in intestinal fluid accumulation and defecation frequency in enteropooling assays, indicate that FN extract might interact with intestinal motility or fluid secretion pathways, providing antidiarrheal benefits. Additionally, the chloroform-soluble fraction of FN exhibited moderate thrombolytic activity, achieving 39.79% clot lysis compared to streptokinase (69.52%). Although this thrombolytic potential is moderate, it indicates that the fraction contains compounds with some fibrinolytic properties, which could be useful as adjunctive agents in preventing thrombosis. ConclusionOverall, the cold methanol extract of FN leaves demonstrates the therapeutic potential in preclinical settings. Future research is warranted to isolate the specific bioactive compounds and explain their mechanisms of action to further support the development of new treatments and contributing to modern medicinal practices based on this plant leaves.

plant biology↗

bioRxiv2024

In silico modelling and characterization of Epstein Barr virus LMP1 protein

Latent membrane protein 1 (LMP1) plays a crucial role in Epstein-Barr virus (EBV)s ability to establish latency and is involved in the development and progression of EBV-associated cancers. Additionally, EBV-infected cells affect the immune responses, making it challenging for immune system to eliminate. Due to the aforementioned reasons, it is important to understand the structural features of LMP1 which is essential for the development of novel cancer therapies that target its signaling pathways. To date, there is no complete LMP1 protein structure therefore in our work, we modeled the full-length LMP1 containing the short cytoplasmic N-terminus, six transmembrane domains and a long-simulated C-terminus. Our model showed good stability and protein compactness evaluated through accelerated-Molecular Dynamics where the conformational ensemble exhibited compact folds, particularly in the transmembrane domains. Our results suggest that specific domains or motifs, predominantly in the C-terminus domain of LMP1 show promise as potential drug targets. As a whole, our work provides insights on key strucutral features of LMP1 that will allow the development of novel LMP1 therapies.

biochemistry↗

bioRxiv2024

Re-evaluation of some popular CRAC channel inhibitors structurally similar but mechanistically different?

Ca2+-release activated Ca2+ (CRAC) channels mediate the store-operated Ca2+ entry (SOCE) that is used almost by all cells to replenish depleted intracellular Ca2+ stores. Aberrant CRAC channel functions have been implicated in various diseases including autoimmune diseases and as such these channels and the resultant pathway (SOCE) have been intensively studied by academic and pharmaceutical scientists over decades. To date, several small molecule inhibitors and few antibodies have been developed against these ion channels. Some small molecule inhibitors of the CRAC channels, namely BTP2, Synta66, Pyr6 and GSK-7975A, are widely used as tool compounds in the field but their mode of action remains unclear. In this study, we used whole cell electrophysiology supported by in silico modelling to propose likely binding site and binding mode of these four popular CRAC channel inhibitors.

biophysics↗

bioRxiv2024

Cav3.1 is a leucine sensor in POMC neurons mediating appetite suppression and weight loss

Hypothalamic leucine sensing promotes satiety and weight loss but an understanding of how leucine regulates neuronal activity is lacking. Here we show that Cacna1g, encoding the T-type voltage-gated calcium channel Cav3.1, is enriched in hypothalamic leucine-sensing neurons and mediates leucine sensing. Pharmacological inhibition of Cav3.1 blunts leucine-induced activation of POMC neurons as well as the anorectic response to leucine in vivo. In addition, genetic deletion of Cacna1g in POMC neurons abolishes the appetite- and weight-suppressive effects of high-protein feeding. Mechanistically, we show that leucine binds to the voltage-sensing segment of Cav3.1, thereby reducing its threshold for voltage-dependent activation. Last, pharmacological activation of hypothalamic Cav3.1 promotes weight loss in diet-induced obese mice and potentiates the weight loss response to GLP-1 receptor agonism. These results reveal that Cav3.1 is a neuronal leucine sensor and a relevant weight loss target.

neuroscience↗

bioRxiv2024

Maptcha: An efficient parallel workflow for hybrid genome scaffolding

BackgroundGenome assembly, which involves reconstructing a target genome, relies on scaffolding methods to organize and link partially assembled fragments. The rapid evolution of long read sequencing technologies toward more accurate long reads, coupled with the continued use of short read technologies, has created a unique need for hybrid assembly workflows. The construction of accurate genomic scaffolds in hybrid workflows is complicated due to scale, sequencing technology diversity (e.g., short vs. long reads, contigs or partial assemblies), and repetitive regions within a target genome. ResultsIn this paper, we present a new parallel workflow for hybrid genome scaffolding that would allow combining pre-constructed partial assemblies with newly sequenced long reads toward an improved assembly. More specifically, the workflow, called Maptcha, is aimed at generating long genome scaffolds of a target genome, from two sets of input sequences--an already constructed partial assembly of contigs, and a set of newly sequenced long reads. Our scaffolding approach internally uses an alignment-free mapping step to build a [<]contig,contig[>] graph using long reads as linking information. Subsequently, this graph is used to generate scaffolds. We present and evaluate a graph-theoretic "wiring" heuristic to perform this scaffolding step. To enable efficient workload management in a parallel setting, we use a batching technique that partitions the scaffolding tasks so that the more expensive alignment-based assembly step at the end can be efficiently parallelized. This step also allows the use of any standalone assembler of choice for generating the final scaffolds. ConclusionsOur experiments with Maptcha on a variety of input genomes, and comparison against a state-of-the-art hybrid scaffolder (LRScaf) demonstrate that Maptcha is able to generate longer and more accurate scaffolds in significantly faster runtimes. For instance, Maptcha produces scaffolds with an NG50 length of 4.8Mbp (compared to 171Kbp by LRScaf) for T. crassiceps, and 81Mbp for Human chr 7 (compared to 4.5Mbp by LRScaf), while reducing the runtime from hours to minutes in several cases. We also performed a coverage experiment by varying the sequencing coverage depth for long reads, which demonstrated the potential of Maptcha to generate significantly longer scaffolds in low coverage settings (1x to 10x).

bioinformatics↗

bioRxiv2024

Comprehensive machine learning boosts structure-based virtual screening for PARP1 inhibitors

Poly ADP-ribose polymerase 1 (PARP1) is an attractive therapeutic target for cancer treatment. Machine-learning scoring functions constitute a promising approach to discovering novel PARP1 inhibitors. Cutting-edge PARP1-specific machine-learning scoring functions were investigated using semi-synthetic training data from docking activity-labelled molecules: known PARP1 inhibitors, hard-to-discriminate decoys property-matched to them with generative graph neural networks and confirmed inactives. We further made test sets harder by including only molecules dissimilar to those in the training set. Comprehensive analysis of these datasets using five supervised learning algorithms, and protein-ligand fingerprints extracted from docking poses and ligand only features revealed two highly predictive scoring functions. The PARP1-specific support vector machine-based regressor, when employing PLEC fingerprints, achieved a high Normalized Enrichment Factor at the top 1% on the hardest test set (NEF1% = 0.588, median of 10 repetitions), and was more predictive than any other investigated scoring function, especially the classical scoring function employed as baseline. Scientific ContributionWe present the first PARP1-specific machine-learning scoring functions for structure-based virtual screening. A particularly rigorous evaluation, including test sets with novel molecules and a much higher proportion of challenging property-matched decoys, reveals the most predictive scoring function for this important therapeutic target. Typically, narrow machine learning analyses would have likely missed this promising PARP1-specific scoring function, which is now released with this paper so that others can use it for prospective virtual screening. Key PointsO_LIA new scoring tool based on machine-learning was developed to predict PARP1 inhibitors for potential cancer treatment. C_LIO_LIThe majority of PARP1-specific machine-learning models performed better than generic and classical scoring functions. C_LIO_LIAugmenting the training set with ligand-only Morgan fingerprint features generally resulted in better performing models, but not for the best models where no further improvement was observed. C_LIO_LIEmploying protein-ligand-extracted fingerprints as molecular descriptors led to the best-performing and most-efficient model for predicting PARP1 inhibitors. C_LIO_LIDeep learning performed poorly on this target in comparison with the simpler ML models. C_LI

bioinformatics↗