Biology subjects

Deb, M.

Publications and source records attributed to Deb, M..

6 recordsLinked to original sources

bioRxiv2026

Braintree: An agentic framework for in-silico vision neuroscience

Vision neuroscience has many of the ingredients needed to conduct experiments entirely in-silico. We need AI systems that can reason over scientific knowledge, generative methods for creating precisely controlled stimuli, image-computable models that can predict neural responses with increasing accuracy, and tools that can analyze and interpret data. Here we take the first step toward making a vision neuroscience experiment end-to-end computable. We introduce Braintree, an agentic framework that can autonomously run multi-experiment studies directly with in-silico models of human visual cortex. Given a hypothesis, Braintree can reason over the current literature, design an experiment, build a stimulus set from algorithmically manipulated or synthesized stimuli, run the experiment on candidate brain encoding models, analyze the responses and then use the results to design even stronger tests. The entire loop runs autonomously and can accept human input at any stage. We evaluated Braintree on three increasingly open-ended scientific problems. First, Braintree reproduced published findings using their original stimuli and then asked whether the same effects could be replicated with entirely new stimulus sets that it constructed. Second, it designed new experiments to find stimulus dimensions that could drive two brain regions differently, even when those regions otherwise had similar response profiles. Third, it searched for experiments that could tell competing models of the same brain region apart. Experiments discovered using just two models also exposed differences across a much broader set of models, suggesting that Braintree had found more general dimensions along which brain models disagree. Braintree can augment the capabilities of a vision neuroscientist as a verifier for existing hypotheses, a test-bed for new ideas, and/or a tool for generating stimuli that can discriminate between competing hypotheses or models of the brain. Braintree shows how AI agents and brain models can work together to make in-silico vision neuroscience experiment computable, allowing vision scientists to explore many more experiment conditions and identify the most informative ones to test on the human brain.

neuroscience↗

bioRxiv2026

Zero-shot evaluations expose structured generalization limits in predictive models of the brain

Artificial neural networks (ANNs) can predict neural responses to natural images with remarkable accuracy, making them promising models of human vision. However, prediction scores alone reveal little about what computations these models have captured or how well they generalize beyond the images on which they are trained. Here, we introduce a principled zero-shot model evaluation framework in which ANN-based encoding models are frozen before being tested on not just held out images but entirely new datasets. Our strongest tests repurpose decades of cognitive neuroscience experiments as diagnostic benchmarks for identifying which computations current models capture and where they fail. To implement this framework we built encoding models of human category-selective regions (FFA, PPA, and EBA) using several pretrained ANNs and evaluated these fixed models across 7 independent fMRI datasets and 31 cognitive experiments from 10 published studies. This framework revealed three broad patterns. First, zero-shot evaluations exposed systematic limits to model predictivity that were largely hidden by standard within-dataset cross-validation. These limits were structured varying across brain regions and model classes and depending strongly on the similarity between the images used to build and test the models. Second, current models reproduced some, but not all, classical findings from cognitive neuroscience, providing a diagnosis of the computations that current models have yet to capture. Finally, models that performed well on prediction tests also tended to succeed at reproducing cognitive neuroscience findings, suggesting that prediction and explanation are closely linked. Unlike prediction scores however, cognitive tests help diagnose where and why models fail. Together, the zero-shot evaluation framework provides a scalable and unified approach for evaluating brain models across prediction and cognitive neuroscience tests, which can help us better understand what current models capture and what still remains to be explained.

neuroscience↗

bioRxiv2025

MOSAIC: A scalable framework for fMRI dataset aggregation and modeling of human vision

Recent large-scale vision fMRI datasets have been invaluable resources to the vision neuroscience community for their deep sampling of individual subjects and diverse stimulus sets. However, practical limitations to the number of subjects, stimuli, and trials that can be collected prevent individual fMRI datasets from reaching the scale necessary for modern modeling approaches and robust conclusions. Here, we introduce MOSAIC (Meta-Organized Stimuli And fMRI Imaging data for Computational modeling), a fMRI dataset aggregation framework designed to leverage the richness of individual datasets for computationally intensive modeling and robust tests of generalization. MOSAIC is composed of eight large-scale vision fMRI datasets totaling 93 subjects, 430,007 fMRI-stimulus pairs, and 162,839 naturalistic and artificial stimuli. A shared fMRI preprocessing pipeline and a filtered test-train split minimizes dataset-specific confounds and test-set leakage when aggregating the datasets. Crucially, additional datasets can be integrated into MOSAIC post hoc, allowing MOSAIC to evolve according to the communitys interests. We use MOSAIC to show that perceptually diverse stimulus sets consistently improve decoding accuracy and stability, carrying implications for future fMRI stimulus set design. We then jointly train brain-optimized encoding models across subjects and datasets to predict fMRI activity of all visual cortex and even the whole brain. In silico functional localizer experiments performed on these digital twin models can recover subject-specific category-selective cortical regions, thereby validating our approach. Together, MOSAIC provides a scalable and community-driven solution to build robust, larger-scale models of human vision.

neuroscience↗

bioRxiv2025

A time-resolved atlas of histone modifications during mitotic entry

Mitotic chromosome formation is essential for faithful chromosome segregation in metazoans. While condensin complexes are critical for the formation of rod-shaped mitotic chromosomes, additional mechanisms--particularly those involving phosphorylation and deacetylation of specific histone residues--have been proposed to contribute a further 2- to 4-fold reduction in mitotic chromatin volume. In this study, we employ high-resolution mass spectrometry to determine the kinetics of histone modifications in cell cultures undergoing a highly synchronous mitotic entry at 2.5-minute resolution. Our analysis reveals three different programmes of histone H3 phosphorylation on T3, S10 and S28. These modifications are consistent with methyl-phos switches regulating the association of readers with chromatin other than at promoters. Mass spectrometry and quantitative ChIP-Seq reveal that H3 T3 phosphorylation is a general marker of heterochromatin and not specifically centromeres as previously suggested. Finally, we show that histone acetylation undergoes only modest changes as rod-shaped chromosomes form during unperturbed mitotic entry. Thus, previously reported reductions in acetylation associated with chromosome formation were apparently attributable to delays in mitotic exit used as part of mitotic synchronisation protocols. The mechanism of condensin-independent chromatin compaction in mitosis remains unexplained.

cell biology↗

bioRxiv2020

ALC1 links chromatin accessibility to PARP inhibitor response in homologous recombination deficient cells

The response to Poly (ADP-ribose) polymerase inhibitors (PARPi) is dictated by homologous recombination (HR) DNA repair mechanisms and the abundance of lesions that trap PARP enzymes on chromatin. It remains unclear, however, if the established role of PARP in promoting chromatin accessibility impacts viability in these settings. Using a CRISPR based screen, we identify the PAR-binding Snf2-like ATPase, ALC1/CHD1L, as a key determinant of PARPi toxicity in HR-deficient cells. ALC1 loss reduced viability of BRCA mutant cells and enhanced their sensitivity to PARPi by up to 250-fold, while overcoming several known resistance mechanisms. ALC1 loss was not epistatic to other repair pathways that execute the PARPi response. Instead, ALC1 deficiency reduced chromatin accessibility concomitant with a decrease in the association of repair factors. This resulted in an accumulation of replication associated DNA damage and a reliance on HR. These findings establish PAR-dependent chromatin remodeling as a mechanistically distinct aspect of PARPi responses, implicating ALC1 inhibition as a new approach to overcome therapeutic resistance in HR-deficient cancers.

cancer biology↗

bioRxiv2019

Down-regulated GDF9 and BMP15 mRNAs are the important determining factors for the unexplained female infertility cases advised for assisted reproductive technology

BackgroundIn India, the potential risk factors influencing female infertility are not well-characterized. Hence, this study was initially aimed to determine the multi-factor risk-predictive models for the infertile females from different Indian ethnics. Afterwards, when treatment was performed for correcting the abnormal demographic and clinical parameters obtained from our initial findings, some of the patients were observed to be treatment-non-responsive. This further recommended assessing the expression pattern of GDF9 and BMP15 mRNAs in those unexplained infertile females. MethodologyThis study included 470 Primary, 209 Secondary infertile and 419 age-matched Healthy females, where baseline demographic and clinicopathological factors were assessed. Additionally, from the treatment-non-responsive patients, laparoscopy and trans-vaginal ultrasound-guided collection of ovarian follicular fluid were done for studying the expression pattern of GDF9 and BMP15 mRNAs using q-PCR. Principal FindingsMid-education level, no contraception usage and duration of infertility for >2 years were the critical risk factors for female infertility (OR>>1; p<0.05), whereas elevated platelet count showed ~10-fold more risk of having secondary infertility and high ESR showed 3.33-fold increased threat of primary infertility. Besides, abnormal MCV, LH/FSH ratio, mild-anaemia and hypothyroidism were the best risk-predictive multi-factor model. After correcting these abnormal clinical factors, 6.8% of the patients were treatment-non-responsive, where elevated ESR, hypothyroidism and anaemia remained uncorrected. Interestingly, in those unexplained cases, GDF9 and BMP15 mRNAs were significantly down-regulated (2-4 fold) irrespective of age groups. Additionally, for predicting female infertility, the cut-off for the expression level of GDF9 and BMP15 mRNAs were estimated -2.09 and -2.75, respectively, with a relatively high sensitivity ([≥]62.5%) and specificity (75%). ConclusionsGDF9 and BMP15 mRNAs may be considered as molecular markers for predicting female infertility. Their down-regulation along with hypothyroidism and anaemia probably led to the treatment-non-responsiveness, thus, those patients may be advised to assisted reproductive technology with supplementation of exogenous GDF9 and/or BMP15.

epidemiology↗