Biology subjects

Wozniak, G.

Publications and source records attributed to Wozniak, G..

3 recordsLinked to original sources

bioRxiv2025

DINOcell: Learning Generalizable Perturbation Effects through Self-Distillation

AO_SCPLOWBSTRACTC_SCPLOWPredicting cellular responses to therapeutics is a promising approach for novel target discovery. However, state-of-the-art computational models designed to predict perturbation effects struggle to generalize and outperform simple baselines. We present DINOcell, a weakly supervised framework that adapts self-distillation to single-cell transcriptomics for predicting perturbation effects. We demonstrate that DINOcell outperforms baselines in predicting the effects of single gene perturbations. Furthermore, DINOcell accurately predicts non-additive effects of combination perturbations, indicating its capacity to model complex genetic interactions. Finally, we show that DINOcell learns representations that capture biological signals and is a promising, generalizable approach for in silico perturbation modeling, providing a valuable tool for accelerating therapeutic target discovery.

bioinformatics↗

bioRxiv2025

Integrative analysis of pooled CRISPR screens provide functional insights into AD GWAS risk genes

Emerging research has implicated Alzheimers disease (AD) pathology with dysregulation of many key pathways in microglia, including lipid transport and metabolism, phagocytosis of plaques, and lysosomal function. However, the exact mechanisms underlying these pathways remain poorly understood. Leveraging high-throughput CRISPR screens to understand the interplay between these pathways may enable novel therapeutic strategies for AD and other neurological diseases. Here, we constructed activation and interference CRISPRa/i libraries targeting 203 genes, 71 of which were identified through neurodegenerative GWAS, and 132 additional genes linked to microglial functions. We used this library to conduct pooled CRISPRa/i screens across a range of functional assays relating to lipid metabolism and lysosomal function using a monocytic cell line, THP-1. We identified a core set of lipid and lysosome mediators and validated a subset in primary macrophages. To gain insights into transcriptional states modulated by these genes we also applied the CRISPRa/i libraries to Perturb-seq, enabling us to capture transcriptomic changes. Through non-negative matrix factorization, we identified five gene programs altered by our perturbation library. We then used an integrative analysis of functional screen data with Perturb-seq data that enabled us to uncover novel functions and genetic relationships between perturbations. This multidimensional resource links genetic perturbations to phenotypes and transcriptional programs, establishing a scalable framework for systematic gene discovery in neurodegeneration and beyond. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=162 SRC="FIGDIR/small/660041v1_ufig1.gif" ALT="Figure 1"> View larger version (47K): org.highwire.dtl.DTLVardef@5b2acborg.highwire.dtl.DTLVardef@156a0b1org.highwire.dtl.DTLVardef@4fdbb8org.highwire.dtl.DTLVardef@e2166c_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOGraphical AbstractC_FLOATNO C_FIG

genetics↗

bioRxiv2019

Modulation of emotion perception and memory via sub-threshold amygdala stimulation in humans

A common human experience is noticing that emotional life events are more vividly remembered than dull ones. Studies show that the amygdala plays a central role in such emotionally driven enhancement of memory. With this in mind, we investigated the effect of electrical brain stimulation of the left human amygdala on performance on an emotional perception and emotional memory task. We randomly applied sub-threshold 50 Hz stimulation to the left amygdala in 10 patients (5 female and 5 male) with intracranial electrodes during the encoding portion of an emotional valence perception and emotional memory task. We found that amygdala stimulation did not affect reported valence for neutral stimuli (non-stimulated group average valence 5.34, stimulated 5.38, p=0.68) but it did affect positively (non-stimulated group average valence 7.31, stimulated 6.70, p=0.004) and negatively (non-stimulated group average valence 2.79, stimulated 3.55, p=0.0002) valenced stimuli in effect reporting both valence categories as more neutral. Furthermore, we found that stimulation did not significantly disrupt memory for neutral stimuli (68% vs. 61% correctly remembered p=0.48) or positive stimuli (87% vs. 70% correct, trend towards significant difference p=0.09) but did for negative stimuli (83% vs. 67% correct, p=0.03). These results suggest that electrical brain stimulation by our parameters likely reversibly inhibits amygdala function disrupting neural networks responsible for emotional perception and memory. This effect may have clinical implications in treatment of certain neuropsychiatric disorders, such as emotional dysregulation and post-traumatic stress disorder.\n\nStatement of significanceThe current study builds and expands on extensive prior research into the function of the human amygdala. It provides the first systematic description in humans of a cognitive change brought about by direct electrical stimulation of the amygdala on perception of emotional valence and emotional memory. The results provide further evidence on the importance of the amygdala in human cognition. Likewise, out method utilized to study the function of the amygdala can be extended to study the function of other brain regions in humans, such as the cingulate. While these results are preliminary and need to be duplicated, we aim to further study the effects of amygdala stimulation on emotional processing including possible therapeutic application for diverse group of neuropsychiatric conditions.

neuroscience↗