HPCg Courses

General information

• On this page, you will find a short description of the courses which are offered by HPCg.
• To register, please visit the IOCB intraweb page and sign up for your desired courses via given course interest form.
• The registered interested attendees will be notified about the room and pre-course preparations by e-mail.
• Course invitations are announced regularly on the IOCB Noticeboard.

List of courses

1. AlphaFold 3/ColabFold: Introduction to AlphaFold 3/ColabFold Basic
2. Aurum:
   • Aurum introduction to cluster usage Basic
   • Aurum tips and tricks for advanced usage Intermediate
3. EasyDock: Customizable and scalable docking tool Basic
4. ImageJ/Fiji:
   • Basic image processing Basic
   • Image processing and analysis Intermediate
   • Colocalization analysis in ImageJ/Fiji Intermediate
5. Linux: Introduction for Aurum users Basic
6. ML & AI: Machine Learning & AI Techniques Intermediate
7. Napari: Python image processing package Basic
8. PyMOL:
   • PyMOL Molecular visualization system Basic
   • PyMOL Python scripting Advanced
9. Python: Introduction to Python scripting Basic
10. RDKit: Cheminformatics in Python using the RDKit library Intermediate
11. StreaMD: Automated molecular dynamics simulations tool Basic

---

Description: This introductory course focuses on using AlphaFold (version 3) and ColabFold pipelines for protein structure prediction and analysis. Attendees will learn the fundamentals of AlphaFold’s deep-learning approach, how to retrieve and prepare sequence data, and how to set up and run prediction jobs both in a free cloud-based environment (Google Colab) and on local hardware or the Aurum HPC cluster. In particular, you will learn how to configure ColabFold so it also runs on Aurum, giving you high-performance resources for larger or more complex protein systems. Additional topics include interpreting and refining predicted structures, assessing the confidence metrics (pLDDT, PAE), and touching upon more advanced topics, such as covalent modifiers or nucleic acid (DNA/RNA) structure prediction.

Who should attend:

• Researchers and students in structural biology, biochemistry, or computational biology who need efficient methods for predicting protein structures.
• Anyone wanting hands-on experience running AlphaFold or ColabFold on an HPC environment (Aurum) or a local workstation.

Learning outcomes:

• Understand the principles behind AlphaFold’s deep-learning approach to protein structure prediction.
• Use ColabFold in different setups, including free Google Colab and the Aurum HPC cluster.
• Submit jobs efficiently to Aurum, handle queue submissions, and choose correct resource allocation.
• Interpret structural outputs using confidence metrics (pLDDT, PAE).
• Use predicted structures for more advanced topics, such as covalent modifiers or nucleic acid (DNA/RNA) structure prediction

Course essentials

• Duration: 1 session (4 hours).
• Prerequisites:
  • Basic familiarity with the Linux command line (e.g., completion of hands-on introduction to Linux and Aurum or equivalent).
  • An Aurum account for HPC usage
  • Basic knowledge of structural biology concepts.
• Minimum participants: 5
• Registration required: Yes

---

Description: This beginner-level course provides a hands-on introduction to the Aurum HPC environment with a focus on job scheduling using Slurm and software management via Spack. Participants will learn how to log in to Aurum, navigate the Linux-based environment, submit and monitor computational jobs, and utilize Spack to install and manage software dependencies. The course also explains basic best practices for parallel job submission and resource allocation on a shared HPC cluster.

Who should attend:

• New HPC users or those transitioning from other clusters to Aurum.
• Researchers who need to run computational jobs, compile custom code, or install specialized software on a shared HPC environment.
• Anyone looking to understand how to manage and optimize job workflows using Slurm and Spack.

Learning outcomes:

• Understand the basics of logging into Aurum, setting up SSH keys, and navigating the file system.
• Learn job submission essentials in Slurm, including creating job scripts, specifying resources (nodes, CPUs, GPUs, memory), and managing job queues.
• Gain familiarity with the Slurm commands (sbatch, squeue, scancel, sinfo) and best practices for efficient scheduling.
• Explore how Spack works for software installation and dependency management, from searching repositories to building and loading modules.
• Acquire practical tips for troubleshooting failed jobs and environment conflicts.

Course essentials

• Duration: 1 session (3–4 hours).
• Prerequisites:
• Basic familiarity with the Linux command line.
• An Aurum account with valid credentials.
• Minimum participants: 5
• Registration required: Yes

---

Topics:

• Advanced Job Management
• Utilizing Slurm Features & Constraints
• Monitoring and Notifications
• File sharing (ACL)
• Compiling own software
• Best Practices and Troubleshooting

Who should attend: Advanced users primarily, all welcome.

Course essentials

• Duration: One session, 4 hours.
• Prerequisites:
• Laptop with administrator privileges. Win10 minimum for Windows users, WSL (Windows subsystem Linux) installed, Aurum account. Basic knowledge of Linux. Basic understanding of Aurum, at least on the level of “Hands-on introduction to Linux and Aurum” course.
• Minimum participants: 5

---

Description: EasyDock is an intuitive, streamlined molecular docking tool that simplifies the process of setting up and executing docking simulations of ligands binding to proteins or nucleic acids (receptor). Participants will learn how to install and configure EasyDock, prepare receptor and ligand files, and use standard scoring functions to evaluate binding poses. The course also covers HPC submission scripts, typical parameter tuning, common pitfalls in docking workflows, and basic visualization of docking results.

Who should attend:

• Researchers and technical staff who want a user-friendly molecular docking tool for routine lead optimization or advanced virtual screening.
• Anyone looking to incorporate automated docking pipelines into their HPC workflow without deep prior knowledge of specialized docking software.

Learning outcomes:

• Install and configure EasyDock for both local and HPC (Aurum) environments.
• Understand file preparation steps for protein/ligand docking (e.g., protonation states, partial charges).
• Utilize HPC job submission scripts to run batch docking studies.
• Interpret docking scores and filter results for promising candidates.
• Visualize docking poses and produce publication-ready images or data tables.

Course essentials

• Duration: 1 session (4 hours).
• Prerequisites:
• Basic command-line skills (Linux).
• Aurum accounts for running large-scale docking jobs.
• Familiarity with chemical structures (receptor/ligand) is helpful but not mandatory.
• Minimum participants: 5
• Registration required: Yes

---

Description: This total beginner-friendly course provides a practical introduction to Fiji (Fiji Is Just ImageJ)—an open-source, Java-based image analysis program. Participants will learn the essential steps for installing Fiji, opening images via Bio-Formats, and navigating its main interface. Key topics include managing memory usage, exploring basic selection tools, working with multiple channels, calibrating image scale, and creating 2D/3D projections from z-stacks. In addition to fundamental image adjustments (brightness/contrast, LUTs, thresholding), attendees will practice ROI-based measurements and line scans to quantify fluorescence intensity, object areas, and other morphological characteristics. Emphasis is placed on reproducible, publication-ready workflows—covering proper file formats, scale-bar creation, and best practices for annotation using overlays. By the end of the course, participants will be able to confidently process and analyze multi-dimensional microscopy datasets, from simple 2D images to 3D time-lapse stacks.

Who should attend:

• Students, researchers, and beginners who need hands-on exposure to core image processing and analysis tasks in Fiji/ImageJ.
• Anyone seeking a structured, practical guide to leveraging open-source image analysis for scientific research.

Learning outcomes:

• Fiji Setup & Interface: Install the program, customize settings (e.g., memory allocation), and navigate the main toolbar.
• Image Handling & Formats: Open a wide range of microscopy file types using Bio-Formats, manage z-stacks, and save/export data in publication-friendly formats.
• Basic Image Processing: Perform noise filtering, background subtraction, and thresholding; adjust brightness/contrast and apply color look-up tables (LUTs).
• Measurement & Analysis: Use region-of-interest (ROI) tools and the ROI manager for manual and semi-automated measurements; conduct line scans and profile plots to quantify intensities or resolve structures (FWHM).
• 3D & Advanced Visualization: Create z-projections, perform orthogonal views and 3D projections, and generate movies for dynamic datasets.

Course essentials

• Duration: 1 sessions (2 hours).
• Prerequisites:
• A laptop with administrator privileges (to install Fiji and any required plugins)
• Minimum participants: 5
• Registration required: Yes

---

Description: Building on the foundational skills covered in the beginner-level workshop, this intermediate course delves deeper into the robust image-processing features of Fiji (ImageJ). Participants will learn how to refine their workflows and tackle more complex image-analysis challenges. Key focus areas include advanced filtering methods (e.g., rolling-ball background subtraction, mean/median filtering), precise thresholding strategies, binary image operations (erosion, dilation, watershed), and automated object segmentation. The course also introduces advanced uses of the ROI (Region of Interest) Manager, automated particle analysis, the Find Maxima tool, and practical applications of the Image Calculator for tasks like masking and ratio-based imaging (e.g., FRET). Through guided, hands-on exercises, attendees will learn best practices in generating reproducible workflows and get comfortable with multi-step pipelines that produce publication-ready data.

Who should attend:

• Researchers, students, and technical staff who already have basic experience in Fiji/ImageJ (or have attended the “Image Processing and Analysis with ImageJ/Fiji for Beginners” course).
• Anyone who needs advanced techniques for segmentation, measurement, and quantitative analysis in microscopy imaging.

Learning outcomes:

1. Advanced Pre-Processing
• Employ advanced filtering techniques (median, Gaussian, rolling-ball) to enhance signal quality.
• Correct uneven illumination and remove background noise.
2. Segmentation & Thresholding
• Apply manual and algorithmic thresholding methods (auto-thresholding, color thresholding).
• Understand binary morphological operations (erode, dilate, open, close) to refine segmented objects.
• Use watershed algorithms to separate touching cells or features.
3. ROI Manager & Particle Analysis
• Generate precise object outlines and manage them with the ROI Manager for multi-channel measurements.
• Apply the Analyze Particles tool for high-throughput counting and feature extraction (e.g., shape descriptors).
4. Advanced Measurement & Analysis
• Use the Find Maxima tool for detecting point-like structures or local intensity peaks.
• Integrate multiple images via the Image Calculator (e.g., ratio-based imaging, mask creation).
• Visualize intensity relationships or co-localization with line scans and custom LUTs.
5. Reproducible Pipelines & Automation
• Combine filtering, segmentation, and measurement steps into robust, reproducible workflows.
• Export and document results (e.g., ROI sets, CSV tables) for publication-quality analyses.

Course essentials

• Duration: 1 sessions (2 hours)
• Prerequisites:
• Basic knowledge of Fiji/ImageJ, ideally attending ImageJ/Fiji Basic Image Processing for Beginners course .
• A laptop with administrator privileges (to install any additional plugins or updates).
• Minimum participants: 5
• Registration required: Yes

---

Description: This intermediate-to-advanced course is designed for researchers who want to deepen their understanding of colocalization analysis using ImageJ/Fiji, drawing on the comprehensive material in Deep Dive into Colocalization Analysis in ImageJ_Fiji.

Participants will learn how to rigorously distinguish true spatial overlap of fluorescent signals from imaging artifacts (e.g., blur, color shifts, crosstalk, and saturation). Topics include best practices for sample preparation, microscope setup, and data acquisition to maximize signal-to-noise while minimizing bleed-through. The workshop then covers key metrics such as Pearson’s and Spearman’s correlation coefficients, Manders’ M1/M2 coefficients, and Manders Overlap Coefficient (MOC), along with methods for setting appropriate thresholds (including Costes’ automatic threshold and randomization tests). Attendees will explore practical workflows using popular Fiji plugins (e.g., BIOP JaCoP, Colocalization Finder, and bUnwarpJ for chromatic shift correction). By the end of the course, participants will be able to confidently design and execute robust colocalization experiments, interpret results with statistical rigor, and communicate findings accurately in publications.

Who should attend:

• Researchers, microscopists, and advanced students with prior experience in ImageJ/Fiji (e.g., comfortable with basic image processing, calibration, and ROI-based measurements).
• Anyone looking to incorporate quantitative colocalization methods into their fluorescence microscopy workflows.

Key Topics: Foundations of Colocalization: Definitions, relationship to molecular interactions, and resolution constraints.Data Quality & Artifact Control: Noise reduction, sample alignment, bleed-through avoidance, color-shift correction, and proper sampling.Quantitative Metrics: Pearson’s vs. Spearman’s correlation; Manders’ M1, M2, and overlap coefficients; thresholds and randomization approaches.Tools & Plugins: Using BIOP JaCoP, Colocalization Finder, and bUnwarpJ for alignment and advanced visualization.Statistical Validation: Costes’ randomization test, handling outliers, and reporting significance.

Practical Exercises: Hands-on image alignment, thresholding strategies, multi-channel data handling, and side-by-side comparisons of colocalization coefficients.

Learning outcomes: By the end of this two-part course, participants will have a solid grasp of how to plan, perform, and critically interpret colocalization analyses in a variety of biological imaging contexts.

Course essentials

• Duration: 1 session (4 hours).
• Prerequisites:
• Familiarity with basic Fiji/ImageJ operations (opening images, brightness/contrast adjustments, ROI Manager, etc.).
• Basic knowledge of fluorescence microscopy is recommended.
• Minimum participants: 5
• Registration required: Yes

---

Description: An introductory Linux/Aurum course.

Topics:

• Overview of Linux system
• command-line basics
• shell scripting (BASH)
• basic administration tasks
• Linux commands to work with cluster.
• Understanding cluster design.
• Accessing Aurum cluster.
• Getting Started with Slurm Documentation & Environment Setup
• Managing Jobs: Submission, Monitoring, and Cancellation

Target: Beginners.

Course essentials

• Duration: One session, 3 hours.
• Prerequisites:
• Laptop with administrator privileges.
• Win10 minimum for Windows users, WSL (Windows subsystem Linux) installed.
• Aurum account.
• Minimum participants: 5
• Registration required: Yes

---

Description: This beginner-friendly yet comprehensive course provides participants with hands-on experience in practical machine learning and artificial intelligence techniques, focusing specifically on image classification and deploying large language models (LLMs) locally. Attendees will gain the necessary skills to run pre-trained LLMs on personal hardware and learn essential methods to retrain or fine-tune deep learning models for specific image classification tasks. Core topics include understanding basic neural network architectures, data preparation and augmentation, training and validation strategies, and leveraging existing open-source models. Participants will practice techniques for fine-tuning image classifiers, deploying state-of-the-art LLMs (e.g., LLaMA) GPUs/CPUs.

Who should attend:

• Researchers and students interested in practical applications of AI, especially image classification and local deployment of large language models.

Learning outcomes:

• Setup & Deployment: Configure your local hardware environment (CPU/GPU) to run pre-trained large language models.
• Data Handling & Preparation: Acquire, preprocess, and augment datasets effectively for robust image classification, understanding best practices for training-validation splits and data normalization.
• Neural Network Basics: Understand core architectures like convolutional neural networks (CNNs), Transformer models, and their applications in image classification and language tasks.
• Model Fine-tuning & Retraining: Execute transfer learning techniques for image classification, fine-tune existing CNN models, and optimize hyperparameters.
• Practical LLM Implementation: Download, deploy, and interact with LLMs locally, including strategies for efficient inference, model quantization, and troubleshooting common deployment issues.

Course essentials

• Duration: 2 sessions (2 hours each).
• Prerequisites:
• A laptop with administrator privileges
• Aurum account
• Familiarity with Python programming
• Minimum participants: 5
• Registration required: Yes

---

Description: Napari is a fast, interactive, multi-dimensional image viewer for Python that is gaining popularity in scientific imaging. This introductory course will guide participants through the basics of installing and configuring Napari, loading diverse image data (e.g., microscopy images, multi-dimensional data), and using built-in tools for visualization and annotation. Attendees will learn how to customize Napari through plugins and integrate it with popular Python libraries such as NumPy, SciPy, and scikit-image for efficient processing pipelines. Emphasis will be on hands-on practice for common image-analysis tasks: measuring objects, manipulating channels, and rapidly prototyping new workflows.

Who should attend:

• Researchers, students, and technical staff who deal with scientific or medical images and want an intuitive Python-based viewer.
• Beginners in Python image processing who need a user-friendly interface plus advanced capabilities for dynamic data exploration.

Learning outcomes:

• Install and set up Napari in a Python environment (conda or system Python).
• Load, visualize, and manipulate 2D/3D/4D images within the Napari viewer.
• Explore basic image processing tasks (thresholding, filtering) using Napari plugins.
• Integrate Napari with Python libraries (NumPy, SciPy, scikit-image) to create automated or semi-automated analysis workflows.
• Save and export images, annotated objects, and measurement results for publication or further analysis.

Course essentials

• Duration: 1 session (2–3 hours)
• Prerequisites:
• Basic familiarity with Python is an advantage.
• Laptop with Anaconda Python environment and admin privileges
• Minimum participants: 5
• Registration required: Yes

---

Description: (Bio)molecular Visualization Software

4-hour course (Theory + Hands-on Session), learn to visualize and manipulate the structures of proteins and ligands and make images thereof

Topics:

• Theory: Interface: GUI - drop-down menus, shortcut buttons, Viewer - 3D models, command line, Object Control Panel, ASHLC menus.
• Practical:
1. View Protein/Ligand Structure: Open Local PDB file/Fetch from PDB
2. Representations: Cartoon, Licorice, Surface
3. Coloring: by a chain, by atom, Background
4. Structural Alignment via Protein
5. Selecting Atoms, Working with Objects
6. Measure distances, hydrogen bonds
7. Mutate protein amino acids
8. Manipulate Ligand Torsions
9. Build ligand modifications
10. Loading NMR models, MD Trajectories
11. Creating Images
12. Saving Session

Target: Beginners.

Course essentials

• Duration: 1 session of 4 hours
• Prerequisites:
• Laptop with administrator privileges. PyMOL installed (instructions will be provided).
• Minimum participants: 5
• Registration required: Yes

---

Description: 4-hour course (Theory + Hands-on Session), learn to visualize and manipulate the structures of proteins and ligands and make images thereof via scripting

Topics:

• Overview of PyMOL scripting and its advantages
• Setting up PyMOL for scripting
• Using PyMOL commands and functions in Python scripts
• PyMOL selection language and operators.
• Manipulating objects and selections in PyMOL using Python
• PyMOL visualization.
• Developing custom scripts for complex tasks
• Integrating PyMOL scripts with other scientific software and tools.

Target: Beginners.

Course essentials

• Duration: 1 session of 4 hours
• Prerequisites:
• Linux or Windows laptop with administrator privileges (Mac users will have troubles but will find it useful too).
• Basic knowledge of PyMOL.
• Ability to read basic Python code.
• PyMOL with Anaconda Python installed (instructions will be provided).
• Minimum participants: 5
• Registration required: Yes

---

Description: Python is a versatile programming language commonly used in all sorts of scientific tasks. It is easy to learn, read, and maintain. Moreover, when combined with Jupyter notebooks, one has the perfect tool for fast prototyping and representationof the results in a single place. Overall, Python is designed to get the job done.

Topics that will be covered:

• Why Python?
• Installation - Conda
• Python as a calculator
• Python types (strings, lists, dictionaries, ...)
• Iteration, indentation, and blocks
• Slicing
• Booleans and truth testing
• Functions

Target: This course is targeted to beginner users. A basic understanding of Python will be required by many courses we will offer.

Course essentials

• Duration:2 sessions of 4 hours.
• Prerequisites:
• Laptop with administrator privileges
• Linux, Mac, Win10 minimum for Windows users
• Minimum participants: 5
• Registration required: Yes

---

Description: The RDKit is a widely used open-source library for cheminformatics. Handling molecules and chemical structures digitally is now an essential component of many workflows, involving tasks such as similarity searching, conformer generation, de novo molecule generation, input preparation for molecular dynamics or QM simulations, featurization of molecules for machine learning and many others. Using Jupyter Notebooks, various features of RDKit will be demonstrated as modular cells that can be reused and adjusted as needed.

Topics that will be covered:

• Loading, saving (from and to standard formats like SMILES, SDF, etc) and manipulating molecules in RDKit
• Visualizing molecules in RDKit
• Calculating descriptors, fingerprints and properties of molecules
• Similarity search between molecules and visualizing chemical space using dimension reduction
• Substructure searching of molecules
• Conformer generation and 3D shape similarity of molecules
• Reaction definition and enumeration of chemical libraries
• Large scale similarity search based on synthons

Who should attend:

• Who want to deal with small or large amounts of molecules on the computer
• Who want to automate parts of their chemical workflows, e.g. automated input preparation for simulations
• Who want to perform classic cheminformatic operations like similarity search, combinatorial enumeration of reaction products or property prediction

Course essentials

• Duration: Duration: 1 session, 4h block.
• Prerequisites:
• Basic Python knowledge (e.g. knowing how to set up and environment, basic notions of syntax, basic notions of datatypes).
• Minimum participants: 5
• Registration required: Yes

---

Description: StreaMD is a specialized workflow for running and analyzing molecular dynamics (MD) simulations of mostly protein-ligand systems. This course introduces participants to setting up MD systems (proteins, membranes, or nucleic acids), preparing input files, and managing extended simulations on HPC resources with real-time data capture.

Who should attend:

• Researchers familiar with the fundamentals of molecular dynamics
• Anyone who wants

Learning outcomes:

• Configure StreaMD workflows to run and monitor simulations in real time on Aurum.
• Prepare systems (proteins, membranes, ligands using AmberTools and run using GROMACS).
• Automate analysis steps (RMSD, RMSF, hydrogen bonds, interaction fingerprints, MM-GBSA interaction energy)
• Identify and troubleshoot common pitfalls in HPC-based MD simulations.
• Use streaming data to guide simulation extensions, parameter adjustments, or dynamic analyses.

Course essentials

• Duration: 1 session (4 hours).
• Prerequisites:
• Basic knowledge of MD principles (e.g., from a general MD/biophysics course).
• Familiarity with Linux command-line usage and HPC cluster job submission.
• Aurum account for HPC resources.
• Minimum participants: 5
• Registration required: Yes

---

Personnel

For inquiries or course suggestions, contact Petro Khoroshyy.

Petro Khoroshyy

Course organization