Bioinformatics Training & Education Program

Biowulf is the NIH high-performance computing (HPC) cluster supporting biomedical research across the Intramural Research Program.

In this special opening seminar, members of the Biowulf team will discuss the history, current capabilities, and future directions of this critical NIH resource. Whether you are an experienced user or new to HPC, this talk will provide valuable insight into the system that powers so much NIH science.

Streamlining Data Sharing: Practical Tools and Researcher Stories from the NIH GREI

A clear and comprehensive Data Management and Sharing (DMS) Plan is essential for meeting NIH policy requirements. This session introduces GREI’s guide to help you incorporate generalist repositories into your DMS Plan (https://doi.org/10.5281/zenodo.14278957), offering recommended language and concrete examples. Learn how to write a stronger, more compliant plan and hear stories from researchers benefiting from sharing data via GREI repositories.

This intermediate Quarto lesson expands beyond the basics to focus on document control and reproducible reporting workflows. Participants will learn to manage document structure with YAML, create cross-references and citations, use and customize themes, and leverage Quarto shortcodes and extensions. Time permitting, we will also explore adding interactivity with Shiny. A basic understanding of Markdown and familiarity with what Quarto is and how it works is recommended.

Scientific discovery is increasingly limited not by data availability, but by our ability to integrate evidence, generate hypotheses, and iteratively test them at scale. Recent advances in foundation models and large language models suggest a new paradigm: AI systems that not only model data, but actively participate in the scientific process as agents. In this talk, I will present a unified view of our recent work on foundation models and agentic systems that aim to make biomedical knowledge transferable, multi-scale, and scientifically testable.

First, I will discuss Universal Cell Embeddings (UCE), a self-supervised foundation model that produces robust, annotation-free cell representations that generalize across datasets and species, enabling zero-shot transfer for single-cell biology without per-dataset retraining. Building on this “universal” cell representation layer, I will introduce PULSAR, a multi-scale, multicellular architecture that explicitly propagates information from genes to cells to multicellular systems, yielding unified donor-level representations for tasks such as disease classification, biomarker prediction, and forecasting future clinical events in the human immune system.

Second, I will connect these models to the broader agenda of the AI Virtual Cell: high-fidelity, multi-scale neural simulators of cellular state and dynamics, and the key scientific and engineering priorities needed to make them real and useful for biology and medicine. Finally, I will move from models to agents. Biomni defines a general-purpose biomedical agent environment with a large, structured action space grounded in real biomedical tools, software, and databases—enabling LLM-based agents to do biomedical work, not just talk about it. To ensure that agent-generated claims can be validated rigorously, I will present Popper, an agentic hypothesis-validation framework inspired by falsification, combining LLM-driven experimental design with sequential statistical testing and explicit Type-I error control. Together, these systems suggest a path toward AI that learns universal biological representations, composes them across scales, and supports end-to-end discovery loops grounded in tools, data, and statistical rigor.

If you use AI tools even occasionally, you’ve probably spent more time than you’d like rewriting prompts, tweaking outputs, or trying to remember “that one prompt that worked.” This live, hands-on class shows you how to stop starting over. You’ll learn how to turn your best prompts into reusable, high-quality assets—stored and shared using the Microsoft 365 tools you already work in every day. In under two hours, you’ll learn practical prompt design techniques that work across tools like ChatGPT, Claude, and CHiRP, and how to organize them in Teams, SharePoint, Word, Excel, and Loop so they’re easy to find, reuse, and improve. The focus is real NIH work, responsible AI use, and immediately applicable skills. You’ll leave with ready-to-use templates, example prompts, and a clear system you can apply the same day to save time, improve results, and make AI a reliable part of your workflow—not an experiment you have to rethink each time.

ChatGPT 101 training is part 1 of a three-part series.  

This one-hour online training led by OpenAI experts will cover the fundamentals of using ChatGPT Enterprise effectively in your daily NIH workflows. Attendees will learn to navigate the ChatGPT interface, implement practices for prompt writing, and utilize key features, such as working with files, search functions, and content drafting in Canvas. The training will also demonstrate real-world use cases for improving productivity and highlight security and compliance features tailored for NIH staff. 

By the end of this training, attendees will be able to:  

• Use ChatGPT Enterprise’s foundational features, including Working with documents, Search, and Canvas. 
• Apply effective prompt strategies to generate accurate, useful outputs for NIH-specific tasks. 
• Understand best practices to help ensure responsible use of generative AI tools like ChatGPT. 

Attendees are not expected to have any prior knowledge of the tool to be successful in this training. 

Qlucore Omics Explorer is a desktop-based point-and-click software with built-in machine learning capabilities. It enables RNA sequencing (bulk and single cell), proteomics and metabolomics analysis. This software is available for NCI CCR scientists upon submitting a ticket at https://service.cancer.gov/ncisp. In this demonstration-only class, Qlucore scientist will illustrate bulk RNA sequencing analysis workflow starting from expression table import through performing normalization, differential expression and pathway analysis, and creating visualizations. Experience using or installation of this software is not required for attendance. Participation is restricted to NIH staff.

This 45-minute online Lunch and Learn training will help attendees develop their own customized strategy for responsibly incorporating generative artificial intelligence (AI) tools, such as ChatGPT, into their workflows. 

By the end of this training, attendees will be able to: 

• Assess appropriate use cases for generative AI tools within their specific research/work context 

• Develop a customized generative AI usage strategy 

• Document their approach for using generative AI tools 

Attendees are not expected to have any prior knowledge of generative AI tools to be successful in this training. 

ChatGPT 102 training is part 2 of a three-part series.  

This one-hour online training led by OpenAI experts will dive deeper into intermediate features and strategies for maximizing ChatGPT Enterprise in NIH workflows. Building on the fundamentals from ChatGPT 101, this training will focus on intermediate features including Custom GPTs, Projects, Data Analysis, coding in Canvas, and Deep Research to enable broader value creation and collaboration with ChatGPT. Attendees will also learn how to integrate ChatGPT into specialized tasks and optimize outputs for NIH-specific use cases. 

By the end of this training, attendees will be able to: 

• Create and customize GPTs and projects to serve as tailored assistants for NIH-specific initiatives and domains. 
• Utilize additional intermediate features including Data Analysis, coding in Canvas, and Deep Research, to handle complex tasks and collaborative workflows. 
• Implement best practices for integrating ChatGPT into broader NIH processes while maintaining compliance and security standards. 

Attendees are expected to be familiar with the basic functions of ChatGPT to be successful in this training (gained by attending ChatGPT 101), attending another relevant training, and/or using ChatGPT previously).  

This one-hour online training, is the first of a two-part series, which introduces participants to cleaning and exploring a patient health dataset using Python and pandas. Attendees will load tabular data, inspect structure and data types, summarize columns, and identify common data quality problems such as missing values, inconsistent formats, and duplicate records. They will then apply practical fixes, including standardizing height and weight units, parsing and normalizing dates of birth, splitting combined fields, and using Boolean masks to flag or correct implausible values.​

By the end of this session students will be able to:

• Import CSV data into pandas DataFrames and quickly understand column types, basic statistics, and overall data quality.​
• Identify duplicate or repeated patient records and decide whether to keep, correct, or remove them.​
• Detect and handle missing or inconsistent values using methods such as isna, fillna, filtering, and conditional replacement.​
• Standardize mixed formats (for example, heights with and without units, date strings in different formats, and numeric values embedded in text).​
• Create derived columns such as systolic and diastolic blood pressure, and use logical conditions to flag questionable or out-of-range values.​

Attendees are expected to have:

• Basic Python coding knowledge
• Familiarity with an IDE and loading script and data files into the IDE. (Colab, Jupyter Notebooks) 

Requirements: 

• Participants will receive a script file and data files prior to the training. These should be loaded and ready to use before the training session begins. 

You can register for Part 2 in this series via the link below: 

https://www.nihlibrary.nih.gov/training/introduction-data-wrangling-using-python-part-2-2

This one-hour online training, the second session of the two-part series,  focuses on reshaping and enriching the cleaned patient dataset to prepare it for analysis and reporting. Attendees will practice splitting and recombining columns (for example, separating full names into first and last names), converting columns to appropriate data types, and engineering new fields such as outlier indicators and blood pressure status labels. The session also covers merging multiple tables (patient details, contact information, and subsets of records) and filtering or subsetting data to answer specific analytical questions.​

By the end of this training, attendees will be able to:

• Reshape and restructure data by splitting and combining columns, changing data types, and reordering or selecting relevant fields.​
• Engineer clinically useful features, including z-score–based outlier flags, hypertension indicators, and combined status columns for downstream models or dashboards.​
• Merge and join DataFrames using common keys (such as patient ID) to bring together core data with supplemental tables like contact information.​
• Filter and subset records based on multiple conditions (for example, patients with diabetes and abnormal blood pressure) to create analysis-ready datasets.​

Attendees are expected to have:

• To have attended Intro to Data Wrangling Using Python - Part 1 of the series
• Basic Python coding knowledge

Familiarity with an IDE and loading script and data files into the IDE. (Colab, Jupyter Notebooks) 

Requirements: 

• Participants will receive a script file and data files prior to the training. These should be loaded and ready to use before the training session begins. 

You can register for Part 1 in this series via the link below: 

https://www.nihlibrary.nih.gov/training/introduction-data-wrangling-using-python-part-1-2

In partnership with the NIH Clinical Center's Biostatistics and Clinical Epidemiology Service (BCES), the NIH Library is offering several trainings that cover general concepts behind statistics and epidemiology. These trainings will help participants better understand and prepare data, interpret results and findings, design and prepare studies, and understand the results in published literature. 

This three-hour online training will provide a review of study designs in biomedical research. This training will also cover details related to case studies/series, ecological, cross-sectional, case-control, and cohort studies, clinical trials, and other study designs and considerations. Time will be devoted to questions from attendees and references will be provided for in-depth self-study. 

By the end of this training, attendees will be able to:  

• Describe two broad categories of study designs 

• Provide examples of descriptive and analytic studies 

• Explain the advantages and disadvantages of analytic studies 

• Understand the differences between observational and experimental studies 

• List other types of atypical study designs 

This 45-minute online training provides a high-level overview of recent developments in artificial intelligence (AI). Each session highlights emerging trends, tools, and use cases in the evolving AI landscape, with an emphasis on practical relevance and responsible use. Whether you're just getting started or looking to stay current, this training offers timely insights in a concise format.  

By the end of this training, attendees will be able to:   

• Summarize key trends and developments in AI 

• Identify new tools, capabilities, or applications relevant to their work 

• Describe considerations for ethical and responsible use of AI technologies 

Attendees are not expected to have any prior knowledge to be successful in this training. 

Advanced ChatGPT training is part 3 of a three-part series. 

This one-hour online training, led by OpenAI experts, is for those who have completed the ChatGPT 101 and 102 trainings. The training will focus on leveraging two of ChatGPT Enterprise's most powerful features: Custom GPTs and Data Analysis. Attendees will learn how to create specialized GPTs tailored for specific NIH tasks and how to use the Data Analysis feature to upload, interpret, and visualize data sets for deeper insights. This training is designed to provide the skills needed to apply these advanced tools to complex, enterprise-level projects. 

By the end of this training, attendees will be able to: 

• Build and deploy Custom GPTs tailored to specific NIH workflows. 
• Use the Data Analysis feature to upload, analyze, and visualize data. 
• Apply advanced techniques to solve complex problems using ChatGPT Enterprise. 

Attendees are expected to be able to utilize ChatGPT to be successful in this training.  

You can register for the other trainings in this series via the link(s) below:  

 ChatGPT 101

ChatGPT 102

In this talk, Dr. Carey will describe how Bioconductor approaches new challenges in supporting open method development and reproducible
analyses in genomic data science. He will discuss aspects of the project that bear on education in cancer epidemiology and
computational cancer genomics, and on emerging topics in software and data engineering for scalable omics analyses.