Defragmentation Training School 2 - NEUBIAS Academy & EOSC-Life 2023

A training school for the new generation of bioimage analysts. Topics: workflow-based image analysis and new integrated methods for cloud computing applied to life sciences.

"Defragmentation TS2: bringing BioImage Analysts to the cloud!”

Programme:

Date	Day	Sessions	Speakers	Time
8 May	Day 1	Welcome. Overview and logistics Student introductions Introduction: What is bioimage analysis? Jupyter for interactive cloud computing Cloud-hosted image data storage, visualisation and sharing Work on your own data OMERO	Paula Sampaio & Rocco D'Antuono Kota Miura Guillaume Witz Bugra Oezdemir Rocco D'Antuono & other Trainers Petr Walczysko	09:00 09:30 11:30 14:00 15:30 16:00
9 May	Day 2	Parallelization and heterogeneous computing: from pure CPU to GPU-accelerated image processing Zero code Deep Learning tools for Bioimage Analysis Analysis of Microtubule Orientation CellProfiler for HCS data on the cloud Work on your own data	Marcelo Zoccoler Daniel Sage and Estibaliz Gomez de Mariscal Thomas Pengo Nodar Gogoberidze & Anna Klemm Rocco D'Antuono & other Trainers	09:00 11:30 14:00 15:00 17:00
10 May	Day 3	Metrics and Benchmarking BIAFLOWS: A BioImage Analysis workflows benchmarking platform Machine and Deep Learning on the Cloud: Classification Machine and Deep Learning on the Cloud: Segmentation Work on your own data	Martin Maška Sébastien Tosi, Volker Baecker, and Benjamin Pavie Felix Mercier Ignacio Arganda Carreras Rocco D'Antuono & other Trainers	09:00 10:00 12:00 12:30 16:30
11 May	Day 4	Symposium
12 May	Day 5	Symposium

Defragmentation TS2 is part of the NEUBIAS 5th Conference: https://eubias.org/NEUBIAS/neubias2020-conference/portugal-2023/

Scientific Organizers: Paula Sampaio, Rocco D'Antuono, Aastha Mathur, Beatriz Serrano Solano, Ana Stojiljkovic, Maria Azevedo, Kota Miura, Julien Colombelli.

Event Timing: 8-12 May 2023, Porto, PT

Defragmentation TS2 Contacts:

Paula Sampaio, i3S, Porto, sampaio@i3s.up.pt

Rocco D'Antuono, The Francis Crick Institute, London, rocco.dantuono@crick.ac.uk

Please use the following subject: Defragmentation TS2

---

Training School preparation

Trainees are kindly asked to get ready for the TS2 by doing the following homework in advance.

Day 1

Get the following accounts for :

• Galaxy account: https://live.usegalaxy.eu/login/start?redirect=None

• Google account for Colab : Google Sign up (a working Google account, not a linked account that you manage with Google). Make sure the account has enough free space (~ several GB) on it.

• A Github account for Binder: Sign up

• follow the instructions from Dr. Bugra Oezdemir to access BAND and run the script that installs the needed software on the remote desktop.

  • Instructions: https://tinyurl.com/3rcwhs4d
  • Repository for the tools: https://github.com/Euro-BioImaging/OME_Zarr_Tools
  • Training page: https://neubias.github.io/training-resources/ome_zarr/index.html

Day 2

GPU-Accelerated Image Processing and Machine Learning

• Download and install mambaforge following the instructions in this blog post: https://biapol.github.io/blog/mara_lampert/getting_started_with_mambaforge_and_python/readme.html Follow the instructions there up to the 'Starting napari' sub-title.

• Link to slides: https://f1000research.com/slides/12-473

• Link to Github repository with exercises/demonstrations: https://github.com/zoccoler/GPU_Accelerated_Image_Processing_on_Cloud_NEUBIAS_Defragmentation_TS2_2023

Zero-Code Deep Learning tools for BioImage Analysis

Fiji installation:

• Download and install it from here
• Install the plugins deepImageJ and MorphoLibJ using the Updater of Fiji:
  • Help > Update… > Open the “Manage update sites”.
  • Select DeepImageJ, IJPB-Plugins (i.e., MorpholibJ)
• Close Fiji and open it again. You're ready!

Transfer the data to your Google Drive

• Link to the data. Download the ctc-glioblastoma folder.

Open the ZeroCostDL4Mic notebook for the 2D U-Net multilabel

• Link to the notebook in the BioImage Model Zoo.
• You can also find all the notebooks in ZeroCostDL4Mic Wiki Page.

Resources:

• DeepImageJ Gómez-de-Mariscal, E., et al., Nat Meth 2021, doi:10.1038/s41592-021-01262-9
• MorpholibJ Legland, D., Arganda-Carreras, I., & Andrey, P., Bioinformatics, 2016, doi:10.1093/bioinformatics/btw413
• ZeroCostDL4Mic: von Chamier, L. et al., Nat Comm 2022, doi:10.1038/s41467-021-22518-0
• BioImage Model Zoo Ouyang, W., et al., bioRxiv 2022, doi:10.1101/2022.06.07.495102

Instructions for CellProfiler presentation:

If you don't have it yet, install CellProfiler, either:

• built package: https://cellprofiler.org/releases
• python package (in a conda or virtual environment): pip install "cellprofiler==4.2.5"
• from source (also in a new environment): https://github.com/CellProfiler/CellProfiler/wiki
  • this option is more involved, including installing java Download "Beginner Segmentation" materials from here: https://tutorials.cellprofiler.org/

[Optional] install docker from here: https://www.docker.com/, and dockerfile from here: https://github.com/CellProfiler/distribution/tree/master/docker

Day 3:

You need a Docker Hub account with the same name as your GitHub account. Docker Hub allows only letters and digits. ‘-’ on GitHub will be removed for Docker Hub Everything will be made lower-case for Docker Hub, like follows:

volker-baecker -> volkerbaecker

Create you accounts:

• GitHub account: https://github.com/

• Docker Hub account: https://hub.docker.com/

BIAFLOWS sandbox server: https://biaflows-sandbox.neubias.org

---

Training School & Symposium material

Day 1:

• Introduction: What is bioimage analysis?
  Kota Miura
• Jupyter for interactive cloud computing
  Guillaume Witz
• Cloud-hosted image data storage, visualization and sharing
  Bugra Oezdemir
• OMERO
  Petr Walczysko

Day 2:

• Parallelization and heterogeneous computing: from pure CPU to GPU-accelerated image processing
  Marcelo Zoccoler
• Zero code Deep Learning tools for Bioimage Analysis
  Daniel Sage & Estibaliz Gómez de Mariscal
• CellProfiler for HCS data on the cloud
  Nodar Gogoberidze

Day 3:

• Metrics and Benchmarking
  Martin Maška
• BIAFLOWS: A BioImage Analysis workflows benchmarking platform
  Sébastien Tosi, Volker Bäcker & Benjamin Pavie
• Machine and Deep Learning on the Cloud: Classification
  Félix Mercier
• Machine and Deep Learning on the Cloud: Segmentation
  Ignacio Arganda-Carreras

Day 4:

• Session 2: AI4Life
  • Deep learning-based bioimage analysis workflows for all audiences: BiaPy
    Ignacio Arganda-Carreras
  • Deep learning enabled cellular imaging: making it happen for you
    Estibaliz Gómez de Mariscal
  • Segment Anything for Microscopy
    Constantin Pape
  • Deep learning with sparse annotations in cell image segmentation
    Ko Sugavara
  • Data Analysis for Super-Resolution Microscopy in nanomedicine
    Cristina Izquierdo Lozano
  • Arkitekt - Streaming analysis and real-time workflows for microscopy
    Johannes Roos

Day 5:

• Session 3: Data Management and Infrastructures
  • FAIR workflows
    Beatriz Serrano-Solano