Biology subjects

Upadhyay, U.

Publications and source records attributed to Upadhyay, U..

7 recordsLinked to original sources

bioRxiv2026

Conditional Generation And Inpainting Of Non-coding RNA Sequences With Masked Discrete Diffusion

Designing functional non-coding RNA (ncRNA) is fundamental to synthetic biology and RNA therapeutics, yet generative modelling for ncRNA has received far less attention than protein design. We present RNA-MDLM, a framework that extends Masked Discrete Language Models (MDLM) to the conditional generation and inpainting of ncRNA. We make two additions: first, conditioning on RNA-type representations from a pretrained RNA language model, and second, a modified classifier-free guidance scheme (Mod-CFG) that interpolates among conditional, unconditional, and random-sequence probabilities for better control. We also introduce REPAINT GAMES, a benchmark of seven structured masking tasks to probe a model's performance on sequence patterns, structural motifs, and base-pairing per RNA type. Our model is trained on 4.6 million ncRNA sequences spanning six evaluable classes. It produces sequences whose composition and folding statistics closely match natural RNAs. Through extensive ablation studies, we find that the embedding-conditioned model achieves the best balance of structural fidelity, biological novelty, and inpainting accuracy, and its class label steers generation far more strongly than a plain label baseline. We further show that a model trained on a smaller, class-balanced subset can appear more realistic mainly by copying abundant natural sequences rather than learning their rules. We also benchmark against a masked-diffusion model and a family-specific VAE on ribozyme families, and find that a type-conditioned model like ours and a per-family model are solving different tasks, which must be accounted for in a fair comparison. We will release the code and the trained models.

biophysics↗

bioRxiv2025

NucleicBERT: Deciphering the language of nucleic acids

The vast majority of the human genome comprises non-protein-coding regions whose structural and functional roles remain poorly understood. Many of these regions function through RNA, yet progress in deep learning for RNA has lagged behind proteins because most methods rely on abundant structural labels or evolutionary alignments, both sparse for RNA. To address these challenges, we developed NucleicBERT, a self-supervised masked-language model that learns contextual representations capturing local and distal dependencies without requiring alignments or evolutionary information. Explainable AI analysis reveals that the model clusters RNA types in latent space and attends to structural properties like secondary structure and tertiary contacts, effectively "rediscovering" RNA biology from sequence correlations alone. When fine-tuned for downstream structural and functional tasks, NucleicBERT requires only single sequences, yet surpasses current state-of-the-art RNA models. This alignment-free framework addresses the scarcity of annotated 3D RNA data while providing a rapid, computational complement to experimental techniques. By bridging abundant unlabeled primary sequence corpora with more scarce structural annotations, NucleicBERT advances RNA structure prediction and provides insights into the working of LLMs. NucleicBERT is available at https://github.com/KIT-MBS/NucleicBERT.

biophysics↗

bioRxiv2025

Fetal Health Classification Based on CTG

This research paper explores the application of advanced machine learning techniques for fetal health detection using cardiotocography (CTG). Cardiotocography is a pivotal tool in monitoring fetal and maternal health during pregnancy, providing crucial insights into fetal heart rate patterns and uterine contractions. In this work, various predictive models, including logistic regression, nearest neighbors, and gradient boosting classifiers, to analyze CTG data were implemented. The findings indicate that these models can effectively classify fetal health status, with gradient boosting demonstrating the highest predictive accuracy. This work highlights the potential of integrating machine learning methodologies into clinical practice to enhance fetal monitoring, ultimately improving maternal and fetal health outcomes.

bioengineering↗

bioRxiv2024

NucleoSeeker - Precision filtering of RNA databases to curate high-quality datasets

The structural prediction of biomolecules via computational methods complements the often involved wet-lab experiments. Un-like protein structure prediction, RNA structure prediction remains a significant challenge in bioinformatics, primarily due to the scarcity of annotated RNA structure data and its varying quality. Many methods have used this limited data to train deep learning models but redundancy, data leakage and bad data quality hampers their performance. In this work, we present NucleoSeeker, a tool designed to curate high-quality, tailored datasets from the Protein Data Bank (PDB) database. It is a unified framework that combines multiple tools and streamlines an otherwise complicated process of data curation. It offers multiple filters at structure, sequence and annotation levels, giving researchers full control over data curation. Further, we present several use cases. In particular, we demonstrate how NucleoSeeker allows the creation of a non-redundant RNA structure dataset to assess AlphaFold3s performance for RNA structure prediction. This demonstrates NucleoSeekers effectiveness in curating valuable non-redundant tailored datasets to both train novel and judge existing methods. NucleoSeeker is very easy to use, highly flexible and can significantly increase the quality of RNA structure datasets.

bioinformatics↗

bioRxiv2024

Functional Dichotomy of Orbitofrontal Cortex and Anterior Cingulate Cortex Subregions in Decision-Making and Brain-Body Regulation

Mood disorders are associated with complex disruptions in brain networks, including those associated with the orbitofrontal cortex (OFC) and pregenual anterior cingulate cortex (pACC). The functional contribution of the caudal OFC (cOFC) has remained largely unexplored. We investigated the functions of the cOFC and pACC in macaques performing an approach-avoidance task by the combination of multimodal recordings and electrical microstimulation (EMS) of the cOFC. We assessed neural, autonomic and behavioral responses. We found the cOFC to be sensitive to both positive and negative stimuli, whereas the pACC was singificantly more active during aversive outcomes. EMS of the cOFC increased avoidance behavior, suggesting a causal role for the OFC subdivision in cost-benefit decision-making. Physiological measurements were positively correlated with behavioral patterns, emphasizing body-brain synchronization during emotionally significant decision-making. We suggest that the cOFC contributes to inducing pessimistic states, thereby making its dysfunction a potential contributor to the etiology of mood disorders.

neuroscience↗

bioRxiv2024

Influence of Contact Map Topology on RNA Structure Prediction

The available sequence data of RNA molecules have highly increased in the past years. Unfortunately, while computational power is still under exponential growth, the computer prediction quality from sequence to final structure is still inferior to the labour intensive experimental work. Although a reliable end-to-end procedure was already developed for proteins since Alphafold2, while its successor AlphaFold3 can also predict RNA, its confidence in particular for novel sequences and folds appear still limited. Another strategy entails two steps: (i) predicting potential contacts in the form of a contact maps from evolutionary data, and (ii) simulating the molecule with a physical force field while using the contact map as restraint. However, the quality of the structure prediction crucially depends on the quality of the contact map. Until now, only the proportion of true positive contacts was considered as a quality characteristic. We propose to also include the distribution of these contacts, and have done so in our recent studies. We observed that the clustering of contacts, as is common for many AI algorithms, has a negative impact on prediction quality. In contrast, a more distributed topology is beneficial. We have applied these findings from computer experiments to current algorithms and introduced a measure of distribution, the Gaussian score.

biophysics↗

bioRxiv2018

Cdc23/Mcm10 Primase Generates the Lagging Strand-Specific Ribonucleotide Imprint in Fission Yeast

The developmental asymmetry of fission yeast daughter cells derives from inheriting \"older Watson\" versus \"older Crick\" DNA strand from the parental cell, strands that are complementary but not identical with each other. A novel DNA strand-specific \"imprint\", installed during DNA replication at the mating-type locus (mat1), imparts competence for cell type inter-conversion to one of the two chromosome replicas. The biochemical nature of the imprint and the mechanism of its installation are still not understood. The catalytic subunit of DNA Polymerase (Pol) has been implicated in the imprinting process. Based on its known biochemical function, Pol might install the mat1 imprint during lagging strand synthesis. The nature of the imprint is not clear: it is either a nick or a ribonucleotide insertion. Our investigations do not support a role of Pol in nicking through putative endonuclease domains but confirm its role in installing an alkali-labile moiety as the imprint. A detailed genetic and molecular analysis reveals a direct role of the Cdc23/Mcm10 primase activity in installing the imprint in cooperation with Pol and Swi1.

genetics↗