Biology subjects

Semenov, A.

Publications and source records attributed to Semenov, A..

6 recordsLinked to original sources

bioRxiv2026

DEEPNET-LLM: Highly accurate structure prediction of unknown metabolites from mass spectra via their molecular neighborhoods

Most molecular features detected by untargeted mass spectrometry-based metabolomics remain unannotated. This so-called 'dark matter' currently comprises more than 90% and likely holds uncharacterized bioactive chemistry. Structure prediction from fragmentation spectra remains an outstanding challenge, because a spectrum does not uniquely determine a structure and the space of candidate structures is mostly unbounded; correspondingly, existing methods rarely return the correct de novo structure as their top prediction. Here we describe De novo Ego network-Enabled Propagation and Neighborhood Extrapolation of Topology using Large Language Models (DEEPNET-LLM) reasoning approach, which uses Molecular Community Networking (MCN) to restrict the search of a query spectrum to its local neighborhood within the global metabolome, and applies AI reasoning to propagate structural information from neighbors to the query. For a collection of 900 authentic standards (2,249 spectra) that have never been publicly released to exclude possibility of data leakage, DEEPNET-LLM returned the correct structure as its top prediction for 84.7% of spectra. Accuracy was independent of precursor mass across a range over which the number of candidate structures grows by 30 orders of magnitude, and was comparable across ion polarity, adducts and in-source fragmentation. On a random subset of the public MassSpecGym benchmark, DEEPNET-LLM reached 85.9% top-1 accuracy excluding stereochemistry. We showcase the use of DEEPNET-LLM with an example of a discovery of tryptamine amidate of a bile acid in mouse gut, confirmed against a synthetic standard. The DEEPNET-LLM approach is orthogonal to existing in silico methods and can be combined with them.

bioinformatics↗

bioRxiv2026

HNSW-MS: Hierarchical Graph Indexing Enables Accurate Real-Time Mass Spectral Similarity Search at Repository Scale

Spectral similarity comparison is the basis of mass spectrometry-based metabolomics, underpinning library matching, molecular network construction, and repository searches such as MASST. Public repositories now exceed billion spectra, and exhaustive pairwise comparison is increasingly limiting. We introduce HNSW-MS, which implements Hierarchical Navigable Small World graph indexing natively for mass spectra. It operates directly on GC-MS and LC-MS/MS spectra without embedding, so scores are identical to those of exhaustive comparison and results are fully reproducible. On the benchmark dataset of 8.4 million LC-MS/MS spectra, HNSW-MS is up to 900-fold faster than linear search, recall@1 and recall@10 achieve over 90%, with the recall-speedup tradeoff tunable at query. Such rapid retrieval allows for any query spectrum to be near-instantaneously integrated into a global molecular network. We demonstrate specific examples of use of global networking for molecular discovery, including uncovering a new pathway of stenothricin biosynthesis and a new dipeptide conjugate bile acid.

bioinformatics↗

bioRxiv2025

Mosaicism of BRAF V600E in Healthy Skin Predicts Neurodegeneration in Histiocytosis

Most histiocytoses harbor an oncogenic somatic alteration activating the MAP kinase pathway, the most frequent of which is BRAFV600E. Targeted therapy with BRAF or MEK inhibitors is highly effective, but neurodegeneration remains a severe complication. Recent data suggest that neurodegeneration may be secondary to mosaicism for BRAFV600E in microglia. In the prospective observational TARGET-HISTIO cohort study, we investigated extra-CNS mosaicism of BRAFV600E in a cohort of 124 adults with histiocytosis. Seeking widespread mosaicism, we analyzed healthy skin biopsies from areas without histiocytosis or melanocytic infiltration using high-sensitivity digital polymerase chain reaction. BRAFV600Ewas detected in healthy skin in 57/67 (85%) patients with BRAFV600Ehistiocytosis and was present in all second and third skin biopsies (n=6). BRAFV600Ewas absent from skin biopsies in all 37 patients with histiocytosis harboring another oncogenic mutation. BRAFV600E was also present in skin biopsies of four /20 patients with histiocytosis of unknown molecular status. Cells harboring BRAFV600E share characteristics with resident dermis macrophages. Only 58% of patients with BRAFV600E in skin also had BRAFV600E in blood leukocytes. Variant allele frequency in skin and blood leukocytes was 16 times lower than in histiocytosis, and frequency in skin was independent of treatment with BRAF/MEK inhibitors. None of the patients with wild type BRAF alleles in skin had neurodegeneration, whereas 22/61 (36%) of patients with BRAFV600E in skin had neurodegeneration. This study demonstrates that mosaicism of BRAFV600Ein healthy skin is frequent in patients with histiocytosis and is strongly associated with neurodegeneration. Key PointsLow allele frequency of the oncogene BRAF V600E is frequent in healthy skin of patients with histiocytosis. Patients without BRAF V600E in skin do not develop neurodegeneration, while 36% of those with BRAF V600E mosaicism do.

cancer biology↗

bioRxiv2024

Aggregation and disaggregation of red blood cells: depletion versus bridging

The aggregation of red blood cells (RBCs) is a complex phenomenon that strongly impacts blood flow and tissue perfusion. Despite extensive research for more than 50 years, physical mechanisms that govern RBC aggregation are still under debate. Two proposed mechanisms are based on bridging and depletion interactions between RBCs due to the presence of macromolecules in blood plasma. The bridging hypothesis assumes the formation of bonds between RBCs through adsorbing macromolecules, while the depletion mechanism results from the exclusion of macromolecules from the inter-cellular space, leading to effective attraction. Existing experimental studies generally cannot differentiate between these two aggregation mechanisms, though several recent investigations suggest concurrent involvement of the both mechanisms. We explore dynamic aggregation and disaggregation of two RBCs using three simulation models: a potential-based model mimicking depletion interactions, a bridging model with immobile bonds, and a new bridging model with mobile bonds which can slide along RBC membranes. Simulation results indicate that dynamic aggregation of RBCs primarily arises from depletion interactions, while disaggregation of RBCs involves both mechanisms. The bridging model with mobile bonds reproduces well the corresponding experimental data, offering insights into the interplay between bridging and depletion interactions and providing a framework for studying similar interactions between other biological cells.

biophysics↗

bioRxiv2024

Ordering molecular diversity in untargeted metabolomics via molecular community networking

Natures molecular diversity is not random but displays intricate organization stemming from biological necessity. Molecular networking connects metabolites with structural similarity, enabling molecular discoveries from mass spectrometry data using arbitrary similarity thresholds that can fracture natural metabolite families. We present molecular community networking (MCN), that optimizes connectivity for each metabolite, rescuing lost relationships and capturing otherwise "hidden" metabolite connections. Using MCN, we demonstrate the discovery of novel dipeptide-conjugated bile acids.

biochemistry↗

bioRxiv2020

The Mehler reaction site is the Phylloquinone within Photosystem I

Photosynthesis is a vital process, responsible for fixing carbon dioxide, and producing most of the organic matter on the planet. However, photosynthesis has some inherent limitations in utilizing solar energy. Up to a third of the energy absorbed is lost in the reduction of O2 to produce the superoxide radical (O2*-), which occurs principally within photosystem I (PSI) via the Mehler reaction. Strikingly, the precise location as well as the evolutionary role of the reaction have long been a matter of debate. For decades, O2 reduction was assumed to take place solely in the distal iron-sulfur clusters of PSI rather than within the two asymmetrical cofactor branches. Here we demonstrate that under high irradiance, O2 photoreduction by PSI takes place at the phylloquinone of one of the branches (the A-branch). This conclusion derives from the light dependency of the O2 photoreduction rate constant, and from the high rates of O2 photoreduction in PSI complexes lacking iron-sulfur clusters and in a mutant PSI, in which the lifetime of this phyllosemiquinone state is extended 100-fold. On these grounds, we suggest that the Mehler reaction serves as a release valve, functioning only when needed, under conditions where both the distal iron-sulfur clusters of PSI and the mobile ferredoxin pool are over reduced. SIGNIFICANCE STATEMENTPhotosynthesis is the process responsible for the oxygenation of the ancient anoxic atmosphere, and the transformation of inorganic carbon to most of the organic matter on Earth. However, it is less commonly appreciated that the appearance of oxygen in the atmosphere led to the unavoidable opposite process in which oxygen is consumed, thereby producing deleterious oxygen radicals such as the superoxide radical. For almost half a decade, the location of the main site of superoxide radical production in chloroplasts has been a matter of debate. We now provide conclusive evidence that it is located in the phylloquinones(s) within photosystem I.

plant biology↗