A5077873393- Advanced Multi-Objective Optimization Algorithms
- Gaussian Processes and Bayesian Inference
- Probabilistic and Robust Engineering Design
- Gene expression and cancer classification
- Genomics and Chromatin Dynamics
- Model Reduction and Neural Networks
- Single-cell and spatial transcriptomics
- Energy Efficiency and Management
- RNA and protein synthesis mechanisms
- Cell Image Analysis Techniques
- Groundwater flow and contamination studies
- Seismic Imaging and Inversion Techniques
- Data Visualization and Analytics
- Spectroscopy and Chemometric Analyses
- Bioinformatics and Genomic Networks
- Soil and Unsaturated Flow
- Simulation Techniques and Applications
- Geothermal Energy Systems and Applications
- Metabolomics and Mass Spectrometry Studies
- Advanced Fluorescence Microscopy Techniques
- Heat Transfer and Optimization
- Epigenetics and DNA Methylation
- Genomics and Phylogenetic Studies
- Spectroscopy Techniques in Biomedical and Chemical Research
- RNA modifications and cancer
Cold Spring Harbor Laboratory
2021-2022
Purdue University West Lafayette
2014-2019
Predictive Science (United States)
2018
Vellore Institute of Technology University
2014
Abstract A problem of considerable importance within the field uncertainty quantification (UQ) is development efficient methods for construction accurate surrogate models. Such efforts are particularly important to applications constrained by high-dimensional uncertain parameter spaces. The difficulty modeling in such systems, further compounded data scarcity brought about large cost forward model evaluations. Traditional response surface techniques, as Gaussian process regression (or...
Abstract Hybrid networks that build upon convolutional layers with attention mechanisms have demon-strated improved performance relative to pure across many regulatory genome analysis tasks. Their inductive bias learn long-range interactions provides an avenue identify learned motif-motif interactions. For maps be interpretable, the layer(s) must identifiable motifs. Here we systematically investigate extent architectural choices in convolution-based hybrid influence motif representations...
Abstract The rapid growth of multi-omics datasets, in addition to the wealth existing biological prior knowledge, necessitates development effective methods for their integration. Such are essential building predictive models and identifying disease-related molecular markers. We propose a framework supervised integration data with priors represented as knowledge graphs. Our leverages graph neural networks (GNNs) model relationships among features from high-dimensional ‘omics set transformers...
The performance of an Earth-Air Heat Exchanger (EAHE) system to be operated in climatic and soil conditions prevailing the Indian district Nagpur is modeled numerically. To do so, a CFD model developed ANSYS Fluent 12.1. validation carried out using data obtained from published literature good agreement established between simulation results experimental data. An earth pipe length 60 m internal diameter 0.1 chosen for validating this validated used further investigate three different lengths...
A problem of considerable importance within the field uncertainty quantification (UQ) is development efficient methods for construction accurate surrogate models. Such efforts are particularly important to applications constrained by high-dimensional uncertain parameter spaces. The difficulty modeling in such systems, further compounded data scarcity brought about large cost forward model evaluations. Traditional response surface techniques, as Gaussian process regression (or Kriging) and...
Abstract Motivation Single-cell Assay for Transposase Accessible Chromatin using sequencing (scATAC-seq) is a valuable resource to learn cis-regulatory elements such as cell-type specific enhancers and transcription factor binding sites. However, identification of scATAC-seq data known be challenging due the heterogeneity derived from different protocols high dropout rate. Results In this study, we perform systematic comparison 7 datasets mouse brain benchmark efficacy neuronal annotation...
The prohibitive cost of performing Uncertainty Quantification (UQ) tasks with a very large number input parameters can be addressed, if the response exhibits some special structure that discovered and exploited. Several physical responses exhibit known as an active subspace (AS), linear manifold stochastic space characterized by maximal variation. idea is one should first identify this low dimensional manifold, project high-dimensional onto it, then link projection to output. In work, we...