Biology subjects

Wang, A. Q.

Publications and source records attributed to Wang, A. Q..

4 recordsLinked to original sources

bioRxiv2026

Rank dependency of rescaled pruning in recurrent neural networks

Throughout development and maturity, neural circuits undergo massive synaptic pruning, yielding highly sparse connectivity while preserving robust population-level computations. These population dynamics are often low-dimensional, allowing task-related computations to be formalized as trajectories within latent subspaces. How such low-dimensional dynamics are preserved amid widespread network sparsification remains unclear. Here, we investigate how different synaptic pruning rules shape low-dimensional dynamics and task performance in recurrent neural networks (RNNs). Moving beyond previous approaches focused on random sparsification of low-rank networks or networks with strictly constrained structures, we systematically evaluate how biologically motivated pruning rules interact with a networks underlying rank. We show that post-pruning dynamics and task performance depend critically on the networks initial rank due to distinct eigenspectral characteristics across rank regimes. Combining mathematical analysis with simulations, we demonstrate that pruning with synaptic rescaling preserves low-dimensional dynamics with minimal distortion in low-rank RNNs, but degrades in the high-rank regime. Our findings suggest that low-rank structure, combined with homeostatic synaptic rescaling, is essential for maintaining stable, low-dimensional dynamics in sparse networks. Author summaryIn biological neural networks, population activity is fundamentally shaped by synaptic connection strengths. During development, many of these connections are eliminated through synaptic pruning--a process thought to minimize metabolic costs and optimize computational efficiency by promoting network sparsity. However, how networks preserve their core functionalities after losing a vast numbers of connections remains an open question. Here, we combine mathematical analysis and simulations to examine how various pruning strategies affect network dynamics and computation in both low- and high-rank recurrent neural networks (RNNs). We show that uncompensated connection removal drastically alters network activity. However, implementing a homeostatic rescaling that strengthens remaining connections preserves original dynamics exclusively in low-rank networks, where connectivity is governed by a few dominant structural patterns. In contrast, pruning with homeostatic rescaling fails to maintain dynamics in high-rank networks. Our findings suggest that low-rank connectivity coupled with network homeostasis is crucial for maintaining brain function throughout significant developmental pruning.

neuroscience↗

bioRxiv2026

Phenotypic screening for small molecules that lower PrP in cultured cells

PrP lowering is a validated therapeutic hypothesis in prion disease. To identify small molecules that reduce PrP levels, we performed phenotypic screening in cultured cells. To prioritize PrP specificity in our primary screen, we generated mouse N2a cells stably expressing GFP and used high content imaging analysis to select compounds that lowered PrP without affecting GFP signal or cell viability. Screening a curated library of 3,492 compounds with annotated mechanisms of action identified two small molecules, EYH (PubChem CID: 71678945) and LCZ (PubChem CID: 24970350), that selectively and dose-dependently lowered PrP. Proteomics on whole cell lysates identified PrP as the #1 or #2 most potently downregulated out of 8,722 proteins detected. Both compounds minimally affected Prnp mRNA, reduced expression of exogenously transfected PrP, and remained potent in non-dividing primary cells, consistent with a post-translational mechanism. Co-treatment with the proteasome inhibitor MG132 yielded accumulation of unglycosylated PrP, demonstrating proteasome clearance of PrP. However, both compounds showed limited or no activity in human cell lines, and failed to reduce PrP in vivo after 14 days of treatment. These findings highlight the challenges associated with mechanism-agnostic phenotypic screening for PrP-lowering compounds and support prioritizing compounds with known mechanisms of action.

neuroscience↗

bioRxiv2025

Constructing Biologically Constrained RNNs via Dale's Backprop and Topologically-Informed Pruning

AO_SCPLOWBSTRACTC_SCPLOWRecurrent neural networks (RNNs) have emerged as a prominent tool for modeling cortical function, and yet their conventional architecture is lacking in physiological and anatomical fidelity. In particular, these models often fail to incorporate two crucial biological constraints: i) Dales law, i.e., sign constraints that preserve the "type" of projections from individual neurons, and ii) Structured connectivity motifs, i.e., highly sparse yet defined connections amongst various neuronal populations. Both constraints are known to impair learning performance in artificial neural networks, especially when trained to perform complicated tasks; but as modern experimental methodologies allow us to record from diverse neuronal populations spanning multiple brain regions, using RNN models to study neuronal interactions without incorporating these fundamental biological properties raises questions regarding the validity of the insights gleaned from them. To address these concerns, our work develops methods that let us train RNNs which respect Dales law whilst simultaneously maintaining a specific sparse connectivity pattern across the entire network. We provide mathematical grounding and guarantees for our approaches incorporating both types of constraints, and show empirically that our models match the performance of RNNs trained without any constraints. Finally, we demonstrate the utility of our methods for inferring multi-regional interactions by training RNN models of the cortical network to reconstruct 2-photon calcium imaging data during visual behaviour in mice, whilst enforcing data-driven, cell-type specific connectivity constraints between various neuronal populations spread across multiple cortical layers and brain areas. In doing so, we find that the interactions inferred by our model corroborate experimental findings in agreement with the theory of predictive coding, thus validating the applicability of our methods.

neuroscience↗

bioRxiv2022

Preclinical Pharmacokinetics and In Vitro Properties of GS-441524, A Potential Oral Drug Candidate for COVID-19 Treatment

Preclinical pharmacokinetics (PK) and in vitro ADME properties of GS-441524, a potential oral agent for the treatment of Covid-19, were studied. GS-441524 was stable in vitro in liver microsomes, cytosols, and hepatocytes of mice, rats, monkeys, dogs, and humans. The plasma free fractions of GS-441524 were 62-78% across all studied species. The in vitro transporter study results showed that GS-441524 was a substrate of MDR1, BCRP, CNT3, ENT1, and ENT2; but not a substrate of CNT1, CNT2, and ENT4. GS-441524 had a low to moderate plasma clearance (CLp), ranging from 4.1 mL/min/kg in dogs to 26 mL/min/kg in mice; the steady state volume distribution (Vdss) ranged from 0.9 L/kg in dogs to 2.2 L/kg in mice after IV administration. Urinary excretion appeared to be the major elimination process for GS-441524. Following oral administration, the oral bioavailability was 8.3% in monkeys, 33% in rats, 39% in mice, and 85% in dogs. The PK and ADME properties of GS-441524 support its further development as an oral drug candidate.

pharmacology and toxicology↗