Analytics and Data Integration (Clinical and Pre-Clinical) Bioinformatics

Key expectations: Develop Predictive Models and Applicable Initiatives, data integration, analytics, machine learning for external and internal data sets, extract data, load data for downstream work - end to end solutions. Clinical data program analytics, processing clinical data sets, utilize for data visualization, data curation. Help the team understand their data.

Organization/Business Function Background

As part of the Data & Computational Science platform in the mRNA Center of Excellence, Research Data Science (RDS) is responsible for leading the application of statistics & advanced data analytics to drive data-driven, decision-making for the advancement of mRNA vaccines and therapeutics. In this context, the RDS team integrates complex and disparate data sources spanning preclinical to translational research, uncovers patterns in the data and develops machine learning models to predict mRNA vaccine performance and further our understanding of the mechanism of action. Familiarity with biology/immunology is preferred.

Scope Description

The project encompasses data integration, statistics, advanced data analytics, modeling, and visualization of diverse test and high-dimensional data from preclinical and clinical studies. Developing efficient and lightweight data science models, methods, or packages may also fall within this scope of work. Developing and constructing extensive bioinformatics workflows that guarantee performance, scalability, and reproducibility in the preprocessing, transformation, and analysis of sizable biological datasets (transcriptomics, flow cytometry, etc.) are some of the primary responsibilities and outputs.

Expected Benefit

The development & implementation of machine learning models to predict clinical performance based on preclinical data is essential for the selection and prioritization of the mRNA-based vaccines to advance to human trials. A core outcome of this work are workflows supporting objective and robust candidate decisions based on predicted immunogenicity and reactogenicity.

Work Product

1. Bioinformatics Workstreams Development

• Design, develop, and optimize end-to-end data pipelines and workflows for preprocessing, transforming and analyzing large-scale biological datasets (transcriptomics, flow cytometry, etc.), ensuring reproducibility, scalability, and performance

• Implement robust data processing components, integrating data quality checks, normalization, feature engineering, and leveraging Python, R, Bioconductor, and relevant bioinformatics toolkits

Deliverables

• End-to-end transcriptomics pipeline from preprocessing to downstream analysis incorporating R/Bioconductor-based bioinformatics approaches for automating normalization, differential gene expression, and geneset enrichment analysis for both individual and groupwise comparisons

2. Clinical Data Programming & Analysis

• Expedite essential preprocessing, analysis, and visualization of clinical data through the development of standardized workflows

• Analyze cross-modal associations to determine correlations between exploratory endpoints and links to outcomes of immunogenicity/reactogenicity

Deliverables:

• Standardized graphics library covering exploratory endpoints (transcriptomics, flow cytometry, serology, etc.) and outcomes (immunogenicity, reactogenicity)

• Associated workflows with well-defined inputs and outputs maximizing code and analysis reusability

3. Data Curation & Exploratory Data Analysis

• Integrate heterogeneous datasets (both internal & external) from multiple modalities (bulk/single-cell RNAseq, flow cytometry, etc.) across studies of interest

• Develop preprocessing and analytical workflows per dataset

• Presentation of exploratory data analysis results to describe key dataset parameters

• Derive features of interest (e.g. differential expression signatures, subject responders vs. non-responders) that can be queried and utilized for cross-study analyses

• Harmonize datasets of interest for utilization in meta-analysis approaches

Deliverables:

• Data warehouse/database of curated key clinical consortia data (Human Immunology Project Consortium/ImmPort) and preclinical studies comprised of multiple modalities

• Per-dataset documentation of parameters and features of interest in an accessible, interactive and searchable dashboard

4. Interactive Visualizations of Multi-Modal Datasets

• Implement responsive and user-friendly interfaces for interactive data visualizations of key preclinical and clinical exploratory endpoints (transcriptomics, flow cytometry, etc.) and outcomes (immunogenicity, reactogenicity)

• Optimize performance and scalability to handle heterogeneous large-scale biological data efficiently

• Provide detailed technical documentation, including architecture diagrams, code documentation, and deployment instructions, to ensure the maintainability and extensibility of the applications

Deliverables:

• Interactive, searchable dashboard serving heatmaps for an atlas of 1500 x 20000 dimensionality that can compute relevant statistics.