A5078524761- Bioinformatics and Genomic Networks
- Cell Image Analysis Techniques
- Gene expression and cancer classification
- Gene Regulatory Network Analysis
- Topological and Geometric Data Analysis
- Single-cell and spatial transcriptomics
- Computational Drug Discovery Methods
- Mathematical Biology Tumor Growth
- Cancer Immunotherapy and Biomarkers
- Cancer Cells and Metastasis
- AI in cancer detection
- Tensor decomposition and applications
- Breast Cancer Treatment Studies
- Algebraic Geometry and Number Theory
- Microtubule and mitosis dynamics
- Radiomics and Machine Learning in Medical Imaging
- MRI in cancer diagnosis
- Data Visualization and Analytics
- Cancer Genomics and Diagnostics
- Genomics and Chromatin Dynamics
- Polynomial and algebraic computation
- Neuroinflammation and Neurodegeneration Mechanisms
- Genetic Associations and Epidemiology
- Ferroptosis and cancer prognosis
- Clusterin in disease pathology
Memorial Sloan Kettering Cancer Center
2017-2023
Allen Institute
2022-2023
Allen Institute for Brain Science
2022-2023
Queen Elizabeth II Medical Centre
2020
The University of Western Australia
2019-2020
Harry Perkins Institute of Medical Research
2019-2020
ARC Centre of Excellence in Synthetic Biology
2020
Menlo School
2014
Dartmouth College
2001
Characterizing cellular diversity at different levels of biological organization and across data modalities is a prerequisite to understanding the function cell types in brain. Classification neurons also essential manipulate controlled ways understand their variation vulnerability brain disorders. The BRAIN Initiative Cell Census Network (BICCN) an integrated network data-generating centers, archives, standards developers, with goal systematic multimodal type profiling characterization....
Abstract We introduce a classification of breast tumors into seven classes which are more clearly defined by interpretable mRNA signatures along the PAM50 gene set than five traditional intrinsic subtypes. Each subtype is partially concordant with one our classes, and two additional correspond to division Luminal B Normal subtypes expression Her2 group. Our class shows similarity myoepithelial mammary cell phenotype, including TP63 (specificity: 80.8% sensitivity: 82.8%), exhibits best...
The mitochondrial inner membrane contains a unique phospholipid known as cardiolipin (CL), which stabilises the protein complexes embedded in and supports its overall structure. Recent evidence indicates that ribosome may associate with to facilitate co-translational insertion of hydrophobic oxidative phosphorylation (OXPHOS) proteins into membrane. We generated three mutant knockout cell lines for CL biosynthesis gene Crls1 investigate effects loss on synthesis. Reduced levels caused...
In the present work, we apply a geometric network approach to study common biological features of anticancer drug response. We use for this purpose panel 60 human cell lines (NCI-60) provided by National Cancer Institute. Our suggests that mathematical tools network-based analysis can provide novel insights into response and cancer biology. adopted discrete notion Ricci curvature measure, via link between robustness established theory optimal mass transport, networks constructed with...
The remarkable growth of multi-platform genomic profiles has led to the challenge multiomics data integration. In this study, we present a novel network-based clustering founded on Wasserstein distance from optimal mass transport. This many important geometric properties making it suitable choice for application in machine learning and clustering. Our proposed method aggregating (aWCluster) is applied breast carcinoma as well bladder carcinoma, colorectal adenocarcinoma, renal lung non-small...
The development of reliable predictive models for individual cancer cell lines to identify an optimal drug is a crucial step accelerate personalized medicine, but vast differences in and characteristics make it quite challenging develop that result high power explain the similarity or drugs. Our study proposes novel network-based methodology breaks problem into smaller, more interpretable problems improve anti-cancer responses lines. For drug-sensitivity study, we used GDSC database 915 200...
Significance A major problem in data science is representation of so that the variables driving key functions can be uncovered and explored. Correlation analysis widely used to simplify networks feature by reducing redundancies, but makes limited use network topology, relying on comparison direct neighbor variables. The proposed method incorporates relational or functional profiles neighboring along multiple common neighbors, which are fitted with Gaussian mixture models compared using a...
Abstract As predictive biomarkers of response to immune checkpoint inhibitors (ICIs) remain a major unmet clinical need in patients with urothelial carcinoma (UC), we sought identify tissue‐based benefit ICIs using multiplex immunofluorescence and integrate these findings previously identified peripheral blood response. Fifty‐five pretreatment 12 paired on‐treatment UC specimens were from treated nivolumab or without ipilimumab. Whole tissue sections stained 12‐plex mIF panel, including CD8,...
In the present work, we study properties of biological networks by applying analogous notions fundamental concepts in Riemannian geometry and optimal mass transport to discrete described weighted graphs. Specifically, employ possible generalizations notion Ricci curvature on manifold spaces order infer certain robustness interest. We compare three (Olivier curvature, Bakry-Émery Forman curvature) some model networks. While exact relationship each definitions one another is still not known,...
Abstract Characterizing cellular diversity at different levels of biological organization across data modalities is a prerequisite to understanding the function cell types in brain. Classification neurons also required manipulate controlled ways, and understand their variation vulnerability brain disorders. The BRAIN Initiative Cell Census Network (BICCN) an integrated network generating centers, archives standards developers, with goal systematic multimodal type profiling characterization....
Abstract We have carried out a topological data analysis of gene expressions for different databases based on the Fermat distance between z scores tissue samples. There is critical value filtration parameter at which all clusters collapse in single one. This healthy samples gapless and smaller than that cancerous ones. After cluster, holes persist larger values Barcodes, persistence diagrams Betti numbers as functions are types cancer constitute fingerprints thereof.
Many biological datasets are high-dimensional yet manifest an underlying order. In this paper, we describe unsupervised data analysis methodology that operates in the setting of a multivariate dataset and network which expresses influence between variables given set. The technique involves geometry employing Wasserstein distance, global spectral form diffusion maps, topological using Mapper algorithm. prototypical application is to gene expression profiles obtained from RNA-Seq experiments...
Abstract We introduce a classification of breast tumors into 7 classes which are more clearly defined by interpretable mRNA signatures along the PAM50 gene set than 5 traditional intrinsic subtypes. Each subtype is partially concordant with one our classes, and 2 additional correspond to division Luminal B Normal subtypes expression Her2 group. Our class shows similarity myoepithelial mammary cell phenotype, including TP63 (specificity: 80.8% sensitivity: 82.8%), exhibits best overall...
For each relative $\operatorname{GL}(V)$-invariant tensor $I\in Λ^{p_1+1}V^{\vee}\otimes .. \otimes Λ^{p_n+1}V^{\vee}$ we construct a weighted differential form $η$ on $(\mathbb{P} V)^{n}$. Then is expressed explicitly with respect to $n$-tuples of frames for tangent spaces at points $\mathbb{P} V$ obtain elements different space. certain invariants $I$, the resulting are shown be multi-focal tensors appearing in machine vision literature (Demazure 1988, Longuet-Higgins 1981, Luong 1992,...
Abstract Many biological datasets are high-dimensional yet manifest an underlying order. In this paper, we describe unsupervised data analysis methodology that operates in the setting of a multivariate dataset and network which expresses influence between variables given set. The technique involves geometry employing Wasserstein distance, global spectral form diffusion maps, topological using Mapper algorithm. prototypical application is to gene expression profiles obtained from RNA-Seq...
Abstract The remarkable growth of multi-platform genomic profiles has led to the multiomics data integration challenge. effective such provides a comprehensive view molecular complexity cancer tumors and can significantly improve clinical out-come predictions. In this study, we present novel network-based method as well clustering technique involving Wasserstein (Earth Mover’s) distance from theory optimal mass transport. We applied our proposed integrative Wasserstein-based (iWCluster)...
Tumorigenesis is a complex process that heterogeneous and affected by numerous sources of variability. This article proposes stochastic extension biologically grounded tumor growth model, referred to as the Norton-Simon-Massagué (NSM) model. First, we study uncontrolled version model where effect chemotherapeutic drug agent absent. Conditions on model's parameters are derived guarantee positivity solution proposed NSM hence its validity describe dynamics volume. The proof makes use...
Application of this framework to text, including editorials and free answers questionnaires as well application (social) survey data, supports their its for these purposes the cartographic methods suggest that micro theories embedded in reduce load a priori assumptions “normally” required both text analysis analysis. The applications research path applicable “ordinary” sociological variables, education, income, occupation, gender, shifts burden argument away from variance explained criteria...
Abstract The emerging field of radiomics, which consists transforming standard-of-care images to quantifiable scalar statistics, endeavors reveal the information hidden in these macroscopic images. This research has found different applications ranging from phenotyping and tumor classification outcome prediction treatment planning. Texture analysis, often reducing spatial texture matrices summary features, been shown be important many latter applications. However, as pointed out studies,...