Biology subjects

Karki, R.

Publications and source records attributed to Karki, R..

8 recordsLinked to original sources

bioRxiv2026

Human-supervised Agentic AI for Hypothesis Generation and Experimental Assistance in Drug Repurposing

Computational drug repurposing has largely been focused on rapid hypothesis generation, yet real-world applications span a far broader lifecycle, from drug candidate suggestion to designing experiments, analyzing assay data, and iteratively refining candidates. Here, we demonstrate that agentic AI can operate throughout this lifecycle. To this end, we developed RepurAgent, a hierarchical multi-agent AI system comprising a supervisor agent and a planning agent that coordinate four specialized sub-agents (research, prediction, data, and report), through a human-in-the-loop design, with episodic memory and retrieval-augmented generation. The system is grounded in data, tools, and standard operating procedures specific for drug repurposing, developed within the REMEDi4ALL consortium. We validated the agentic system across three scenarios spanning the various stages within the repurposing lifecycle: in Acute Myeloid Leukemia, a blinded expert evaluation indicated that RepurAgent produced substantially more novel and mechanistically credible candidates compared to a vanilla LLM baseline; in a retrospective COVID-19 antiviral screen, RepurAgent acted as an adaptive experimental collaborator, prioritizing compounds with AUC-ROC up to 0.99 without predefined thresholds and flagging confounders missed in manual review; and for Multiple Sulfatase Deficiency, it prioritized 81 high-confidence candidates from 5000 compounds, which were further corroborated by domain experts. These results demonstrate that agentic AI can support across the drug repurposing lifecycle, from hypothesis generation to experimental analysis. RepurAgent is open source and deployed at https://repuragent.serve.scilifelab.se/.

bioinformatics↗

bioRxiv2025

Hierarchical small molecule inhibition of MYST acetyltransferases

MYST lysine acetyltransferases (KATs) is a class of epigenetic enzymes critical for cellular function that constitute an emerging therapeutic target in cancer. Recently, several drug-like MYST inhibitors have been reported that show promise in a variety of preclinical models as well as in clinical trials of breast cancer. However, the comparative properties of these small molecules remains to be directly assessed. Here we apply an integrated profiling strategy to systematically define the potency and selectivity of drug-like MYST KAT inhibitors. First, we use optimized chemoproteomic profiling and histone acetylation biormarkers to study the industry-developed KAT inhibitor PF-9363. This reveals dose-dependent engagement of native KAT complexes, with hierarchical inhibition following the order KAT6A/B > KAT7 >> KAT8 > KAT5. Next, we demonstrate how PF-9363s ability to disrupt capture of MYST complex members in chemoproteomic experiments can be leveraged to identify new candidate members of these complexes, including the transcription factor FOXK2. Applying insights from these studies to WM-8014, WM-1119 and WM-3835, which have been extensively applied in the literature as MYST probes, highlights unexpected cross-inhibition and suggests a new framework for how these small molecules and biomarkers may be applied to differentiate KAT6A/B and KAT7-dependent phenotypes. Finally, we benchmark the activity of PF-9363 in the NCI-60 cell line screen, providing evidence that it can inhibit the growth of cell lines that are resistant to other epigenetic inhibitors by engaging the essential MYST enzyme KAT8 at high concentrations. Collectively, our studies indicate the potential for MYST KAT inhibitors to exhibit dose-dependent target engagement reminiscent of kinase inhibitors and specify assays and biomarkers for facile monitoring of selective and hierarchical effects.

biochemistry↗

bioRxiv2024

Lineage-specific proteome remodeling of diverse lung cancer cells by targeted epigenetic inhibitors

Epigenetic inhibitors exhibit powerful antiproliferative and anticancer activities. However, cellular responses to small-molecule epigenetic inhibition are heterogeneous and dependent on factors such as the genetic background and metabolic state of cells, as well as on-/off-target engagement of individual small-molecule compounds. The molecular study of the extent of this heterogeneity often measures changes in a single cell line. To more comprehensively profile the effects of small-molecule perturbations and their influence on heterogeneous cellular responses, we present a molecular resource based on the quantification of chromatin, proteome, and transcriptome remodeling due to histone deacetylase inhibitors (HDACi) in non-isogenic cell lines. Through quantitative molecular profiling of 10,621 proteins, these data reveal coordinated molecular remodeling of HDACi treated cancer cells. HDACi-regulated proteins differ greatly across cell lines with consistent (JUN, MAP2K3, CDKN1A) and divergent (CCND3, ASF1B, BRD7) cell-state effectors. Together these data provide valuable insight into cell-type driven and heterogeneous responses that must be taken into consideration when monitoring molecular perturbations in culture models. We have also built a web interface for the extensive amount of data to allow users to explore the data as a resource for understanding chemical perturbation of diverse cell types.

cancer biology↗

bioRxiv2023

Pharmacophoric-based ML model to filter candidate E3 ligands and predict E3 Ligase binding probabilities

Among the plethora of E3 Ligases, only a few have been utilized for the novel PROTAC technology. However, extensive knowledge of the preparation of E3 ligands and their utilization for PROTACs is already present in several databases. Here we provide, together with an analysis of functionalized E3 ligands, a comprehensive list of trained ML models to predict the probability to be an E3 ligase binder. We compared the different algorithms based on the different description schemes used and identified that the pharmacophoric-based ML approach was the best. Due to the peculiar pharmacophores present in E3 ligase binders and the presence of an explainable model, we were able to show the capability of our ErG model to filter compound libraries for fast virtual screening or focused library design. A particular focus was also given to target E3 ligase prediction and to find a subset of candidate E3 ligase binders within known public and commercial compound collections.

bioinformatics↗

bioRxiv2022

Monkeypox Knowledge Graph: A comprehensive representation embedding chemical entities and associated biology of Monkeypox

SummaryThe outbreak of Monkeypox virus (MPXV) infection in May 2022 is declared a global health emergency by WHO. As of 28th July, 21067 cases have been confirmed and the numbers are on the rise. Unfortunately, MPXV pathophysiology and its underlying mechanisms are not yet understood. Likewise, the knowledge of biochemicals and drugs used against MPXV and their downstream effects is sparse. In this work, using Knowledge Graph (KG) representations we have depicted chemical and biological aspects of MPXV. To achieve this, we have collected and rationally assembled several biological study results, assays, drug candidates, and preclinical evidence to form a dynamic and comprehensive network. The KG is compliant with FAIR annotations allowing seamless transformation and integration to/with other formats and infrastructures. Availability and implementationThe Monkeypox knowledge graph is publicly available at https://github.com/Fraunhofer-ITMP/mpox-kg Supplementary informationSupplementary data are available at Bioinformatics online. ContactReagon.Karki@itmp.fraunhofer.de

systems biology↗

bioRxiv2020

COVID-19 cytokines and the hyperactive immune response: Synergism of TNF-α and IFN-γ in triggering inflammation, tissue damage, and death

The COVID-19 pandemic has caused significant morbidity and mortality. Currently, there is a critical shortage of proven treatment options and an urgent need to understand the pathogenesis of multi-organ failure and lung damage. Cytokine storm is associated with severe inflammation and organ damage during COVID-19. However, a detailed molecular pathway defining this cytokine storm is lacking, and gaining mechanistic understanding of how SARS-CoV-2 elicits a hyperactive inflammatory response is critical to develop effective therapeutics. Of the multiple inflammatory cytokines produced by innate immune cells during SARS-CoV-2 infection, we found that the combined production of TNF- and IFN-{gamma} specifically induced inflammatory cell death, PANoptosis, characterized by gasdermin-mediated pyroptosis, caspase-8-mediated apoptosis, and MLKL-mediated necroptosis. Deletion of pyroptosis, apoptosis, or necroptosis mediators individually was not sufficient to protect against cell death. However, cells deficient in both RIPK3 and caspase-8 or RIPK3 and FADD were resistant to this cell death. Mechanistically, the JAK/STAT1/IRF1 axis activated by TNF- and IFN-{gamma} co-treatment induced iNOS for the production of nitric oxide. Pharmacological and genetic deletion of this pathway inhibited pyroptosis, apoptosis, and necroptosis in macrophages. Moreover, inhibition of PANoptosis protected mice from TNF- and IFN-{gamma}-induced lethal cytokine shock that mirrors the pathological symptoms of COVID-19. In vivo neutralization of both TNF- and IFN-{gamma} in multiple disease models associated with cytokine storm showed that this treatment provided substantial protection against not only SARS-CoV-2 infection, but also sepsis, hemophagocytic lymphohistiocytosis, and cytokine shock models, demonstrating the broad physiological relevance of this mechanism. Collectively, our findings suggest that blocking the cytokine-mediated inflammatory cell death signaling pathway identified here may benefit patients with COVID-19 or other cytokine storm-driven syndromes by limiting inflammation and tissue damage. The findings also provide a molecular and mechanistic description for the term cytokine storm. Additionally, these results open new avenues for the treatment of other infectious and autoinflammatory diseases and cancers where TNF- and IFN-{gamma} synergism play key pathological roles.

immunology↗

bioRxiv2020

The COVID-19 PHARMACOME: Rational Selection of Drug Repurposing Candidates from Multimodal Knowledge Harmonization

The SARS-CoV-2 pandemic has challenged researchers at a global scale. The scientific communitys massive response has resulted in a flood of experiments, analyses, hypotheses, and publications, especially in the field of drug repurposing. However, many of the proposed therapeutic compounds obtained from SARS-CoV-2 specific assays are not in agreement and thus demonstrate the need for a singular source of COVID-19 related information from which a rational selection of drug repurposing candidates can be made. In this paper, we present the COVID-19 PHARMACOME, a comprehensive drug-target-mechanism graph generated from a compilation of 10 separate disease maps and sources of experimental data focused on SARS-CoV-2 / COVID-19 pathophysiology. By applying our systematic approach, we were able to predict the synergistic effect of specific drug pairs, such as Remdesivir and Thioguanosine or Nelfinavir and Raloxifene, on SARS-CoV-2 infection. Experimental validation of our results demonstrate that our graph can be used to not only explore the involved mechanistic pathways, but also to identify novel combinations of drug repurposing candidates.

bioinformatics↗

bioRxiv2020

COVID-19 Knowledge Graph: a computable, multi-modal, cause-and-effect knowledge model of COVID-19 pathophysiology

SummaryThe past few weeks have witnessed a worldwide mobilization of the research community in response to the novel coronavirus (COVID-19). This global response has led to a burst of publications on the pathophysiology of the virus, yet without coordinated efforts to organize this knowledge, it can remain hidden away from individual research groups. By extracting and formalizing this knowledge in a structured and computable form, as in the form of a knowledge graph, researchers can readily reason and analyze this information on a much larger scale. Here, we present the COVID-19 Knowledge Graph, an expansive cause-and-effect network constructed from scientific literature on the new coronavirus that aims to provide a comprehensive view of its pathophysiology. To make this resource available to the research community and facilitate its exploration and analysis, we also implemented a web application and released the KG in multiple standard formats. AvailabilityThe COVID-19 Knowledge Graph is publicly available under CC-0 license at https://github.com/covid19kg and https://bikmi.covid19-knowledgespace.de. Contactalpha.tom.kodamullil@scai.fraunhofer.de Supplementary informationSupplementary data are available online.

bioinformatics↗