Short Courses

Machine learning-and correspondingly, artificial intelligence-has become a dominant technology for data-intensive discovery in nearly all scientific domains. Today, almost all biomedical research employs machine learning techniques to derive new knowledge from complex biological data. This course will introduce the fundamentals of machine learning with a specific focus on the analysis of proteomics data, through a series of lectures and hands-on labs. The primary goal of this course is to promote data literacy for people new to machine learning. Students who complete the course will be equipped to critically evaluate uses of machine learning in scientific literature, recognize common pitfalls, perform basic machine learning analyses, and know when to consult a machine learning expert-and how to communicate with them. The topics that will be covered include: - What is machine learning? - Types of machine learning tasks - The bias-variance trade-off - Machine learning model evaluation - Linear models and logistic regression - Decision tree-based models - Neural networks and artificial intelligence Course participants will: - Recognize when machine learning methods may be beneficial for their research. - Identify common pitfalls in the application of machine learning methods. - Gain confidence to provide constructive feedback for applications of machine learning in the manuscripts they review. - Evaluate the strengths and weaknesses of machine learning approaches presented in the scientific literature. - Gain familiarity with additional resources to deepen their understanding of machine learning. Prerequisites: All participants will need to bring a laptop to perform the lab exercises during the course. A familiarity with proteomics data analysis is recommended. No machine learning experience is needed. No programming experience is necessary; however, some programming experience in any language is recommended and we will introduce Python as part of the course.

This course will cover both theory and practice for analyzing quantitative proteomics data. The course will be a mixture of both "lecture" and "lab" portions. The "lectures" will discuss best practices, considerations, and pitfalls for interpreting label-free and labeled measurements, including those produced by DDA, DIA, PRM, TMT/iTRAQ, and SILAC methods. The "labs" will give students a chance to explore and analyze raw data from different instrument platforms with open-source tools using Windows laptops that they either bring with them or have remote access to. While participants are highly recommended to bring their own computers to get the full experience, participants without Windows laptops will be able to join up as teams to analyze datasets. The goal of this course is to give students a foundation for how data analysis tools work and to build an intuition for what data analysis challenges lurk in their datasets. A portion of the class will be devoted to "office hours" where students will get a chance to discuss their specific quantitative proteomics challenges. Audience: proteomics researchers at all levels that want to learn what common proteomics data analysis tools/methods are doing, how they work, and where they break down.

The lecture focuses on spatial lipidomic strategies in unique clinical samples and how it can be integrated into different multi-omic workflows.

Most proteomics measurements utilize liquid chromatography separations coupled with high resolution/mass accuracy mass spectrometry. With these technologies maturing, it becomes possible to collect data while treating the LC-MS platform as a black box, not fully understanding what's going on inside. This practical lecture-style short course will be divided into three topics: separations, ionization and mass measurement. The separations portion will delve into the basics of LC separations, primarily focusing on reversed phase LC but also briefly touching on other modes. We will discuss sample loading and injection strategies, column dimensions, throughput and flow rate considerations. We will utilize some common free software tools to optimize our separations within the constraints of specific instrument capabilities, and students will be invited to follow along with these software tools on their own as desired. We will then describe electrospray ionization and practical considerations to increase robustness, stability and sensitivity. For the mass spectrometry portion, we will describe the ion path and ion optics encountered in modern mass spectrometers, the (very) basics of mass spectrum interpretation and the mass analyzers commonly encountered in proteomics. The course will be highly interactive such that students will be able to get their own questions answered. The target audience is users of LC-MS-based proteomics technologies who either need a refresher or who have never had formal instruction on the use of the instrumentation.

This course will cover both theory and practice for analyzing quantitative proteomics data. The course will be a mixture of both "lecture" and "lab" portions. The "lectures" will discuss best practices, considerations, and pitfalls for interpreting label-free and labeled measurements, including those produced by DDA, DIA, PRM, TMT/iTRAQ, and SILAC methods. The "labs" will give students a chance to explore and analyze raw data from different instrument platforms with open-source tools using Windows laptops that they either bring with them or have remote access to. While participants are highly recommended to bring their own computers to get the full experience, participants without Windows laptops will be able to join up as teams to analyze datasets. The goal of this course is to give students a foundation for how data analysis tools work and to build an intuition for what data analysis challenges lurk in their datasets. A portion of the class will be devoted to "office hours" where students will get a chance to discuss their specific quantitative proteomics challenges. Audience: proteomics researchers at all levels that want to learn what common proteomics data analysis tools/methods are doing, how they work, and where they break down.

This course will introduce various proteomics techniques for studying stable and transient protein-protein interactions, including affinity purification, proximity labeling, and cross-linking methods. The aim is to provide practical tips on study design, experimental optimization, data analysis, and troubleshooting for protein interaction studies. Protein interaction experiments are often challenged by nonspecific binding, experimental variability, contamination, false discoveries, and data analysis bottlenecks. This course will provide practical strategies to address these challenges and optimize experimental design. Instructors with expertise in affinity purification, proximity labeling, and cross-linking techniques will present real-life examples of problems, common pitfalls, and effective solutions. Through both successful and failed experimental examples, attendees will gain insight into best practices for refining protocols, improving data reliability, and overcoming computational hurdles. An interactive discussion session will also be included, giving attendees the opportunity to share their experience and challenges in protein interaction experiments.