A5079460780- Genomics and Chromatin Dynamics
- Epigenetics and DNA Methylation
- Birth, Development, and Health
- Monoclonal and Polyclonal Antibodies Research
- Renal and related cancers
- Advanced Fluorescence Microscopy Techniques
- RNA and protein synthesis mechanisms
- Cell Image Analysis Techniques
- Protein Structure and Dynamics
- Glycosylation and Glycoproteins Research
- Genomics and Phylogenetic Studies
- Immunodeficiency and Autoimmune Disorders
- vaccines and immunoinformatics approaches
- Biosimilars and Bioanalytical Methods
- Machine Learning in Materials Science
- CAR-T cell therapy research
- Cancer-related molecular mechanisms research
- Transgenic Plants and Applications
- Gene expression and cancer classification
- Renal Diseases and Glomerulopathies
- Genomics and Rare Diseases
- RNA Research and Splicing
- Immune cells in cancer
- Image Processing Techniques and Applications
- Neutrophil, Myeloperoxidase and Oxidative Mechanisms
Stanford University
2024-2025
University of California, Berkeley
2021-2024
Berkeley College
2024
University of California System
2021
Proteins mediate their functions through chemical interactions; modeling these interactions, which are typically sidechains, is an important need in protein design. However, constructing all-atom generative model requires appropriate scheme for managing the jointly continuous and discrete nature of proteins encoded structure sequence. We describe diffusion structure, Protpardelle, represents all sidechain states at once as a “superposition” state; superpositions defining collapsed into...
Deconvolution can be used to obtain sharp images or volumes from blurry encoded measurements in imaging systems. Given knowledge of the system’s point spread function (PSF) over field view, a reconstruction algorithm recover clear image volume. Most deconvolution algorithms assume shift-invariance; however, realistic systems, PSF varies laterally and axially across view due aberrations design. Shift-varying models used, but are often slow computationally intensive. In this work, we propose...
Abstract Generative AI has the potential to redefine process of therapeutic antibody discovery. In this report, we describe and validate deep generative models for de novo design antibodies against human epidermal growth factor receptor (HER2) without additional optimization. The enabled an efficient workflow that combined in silico methods with high-throughput experimental techniques rapidly identify binders from a library ∼10 6 heavy chain complementarity-determining region (HCDR)...
Genomic deep learning models can predict genome-wide epigenetic features and gene expression levels directly from DNA sequence. While current perform well at predicting across genes in different cell types the reference genome, their ability to explain variation between individuals due cis-regulatory genetic variants remains largely unexplored. Here, we evaluate four state-of-the-art on paired personal genome transcriptome data find limited performance when explaining individuals. In...
Discovery and optimization of monoclonal antibodies for therapeutic applications relies on large sequence libraries, but is hindered by developability issues such as low solubility, thermal stability, high aggregation, immunogenicity. Generative language models, trained millions protein sequences, are a powerful tool on-demand generation realistic, diverse sequences. We present Immunoglobulin Language Model (IgLM), deep generative model creating synthetic libraries re-designing...
Abstract Genomic deep learning models can predict genome-wide epigenetic features and gene expression levels directly from DNA sequence. While current perform well at predicting across genes in different cell types the reference genome, their ability to explain variation between individuals due cis-regulatory genetic variants remains largely unexplored. Here we evaluate four state-of-the-art on paired personal genome transcriptome data find limited performance when explaining individuals.
Abstract Background A number of deep learning models have been developed to predict epigenetic features such as chromatin accessibility from DNA sequence. Model evaluations commonly report performance genome-wide; however, cis regulatory elements (CREs), which play critical roles in gene regulation, make up only a small fraction the genome. Furthermore, cell type-specific CREs contain large proportion complex disease heritability. Results We evaluate genomic regions with varying degrees type...
Leading deep learning-based methods for fixed-backbone protein sequence design do not model sidechain conformation during generation despite the large role three-dimensional arrangement of atoms play in conformation, stability, and overall function. Instead, these models implicitly reason about crucial interactions based on backbone geometry known amino acid labels. To address this, we present FAMPNN (Full-Atom MPNN), a method that explicitly both identity each residue, where per-token...
Kidney disease is highly heritable; however, the causal genetic variants, cell types in which these variants function, and molecular mechanisms underlying kidney remain largely unknown. To identify loci affecting we performed a GWAS using multiple function biomarkers identified 462 loci. begin to investigate how affect generated single-cell chromatin accessibility (scATAC-seq) maps of human candidate
Background: A number of deep learning models have been developed to predict epigenetic features such as chromatin accessibility from DNA sequence. Model evaluations commonly report performance genome-wide; however, cis regulatory elements (CREs), which play critical roles in gene regulation, make up only a small fraction the genome. Furthermore, cell type specific CREs contain large proportion complex disease heritability. Results: We evaluate genomic regions with varying degrees...
Abstract Neutrophils are rapidly recruited from the peripheral blood to inflammatory site initiate response against pathogenic infections. The process recruit neutrophils must be properly regulated since abnormal accumulation of can cause organ damage and dysfunction. acute respiratory distress syndrome (ARDS)/acute lung injury (ALI) is a common failure that characterized by infiltration epithelial integrity disruption. Indeed, recent studies suggest role in clinic severity coronavirus...
Abstract Genomic sequence-to-activity models are increasingly utilized to understand gene regulatory syntax and probe the functional consequences of variation. Current make accurate predictions relative activity levels across human reference genome, but their performance is more limited for predicting effects genetic variants, such as explaining expression variation individuals. To better causes these shortcomings, we examine uncertainty in genomic using an ensemble Basenji2 model...
We present a deep-learning method based on Wiener filters and U-Nets that performs image reconstruction in systems with spatially-varying aberrations. train simulated microscopy measurements test experimental data, demonstrating high resolution reconstructions.