CodeJam 2022 - Agenda

.

The 13th CodeJam will again focus on EBRAINS infrastructure themes.

The CodeJam follows the successful model from the previous CodeJams: mainly interactive working sessions of developers and a few talks about general interest topics from the EBRAINS overall computing / service system.

Registration

The registration for the CodeJam is free of charge and requires an EBRAINS account (also free of charge).

Dial-in

Registered attendants (for whom the personal dial-in link has been created) can dial in via this link

(dial in)

• Service for automated executions of Jupyter notebooks (15min)​
• Demo: how to register a new notebook​ (10min)
• Centralized monitoring service (15min)​
• Demo: Monitor system metrics for Debian-based OS​ (10min)

An in-depth Q&A will take place in one of the break-out rooms. ​

(dial in)

Presentation (30min + Q&A): starting point, current status, goal, our definition, experience from Showcase 3 shifting their work into CWL

In parallel:

• in the main room: (Presenters: Sofia Karvounari and Eleni Mathioulaki)
  • Writing Workflows (1hr): TC will give some guidelines and attendees can experiment with writing cwl steps and workflows (Editors and Viewers: https://www.commonwl.org/tools/#editors).
  • Execute Workflows (30min): TC will provide access to VM for execution of workflows. Please contact hbp-tc@athenarc.gr for more information
• break out rooms available for self-organised cooperative coding

(dial in)

• Presentation (30min + questions) - High-level overview of the build and delivery strategy: automated testing and deployment operations, official and experimental release schedule, steps and requirements for integration into the EBRAINS software release

• In parallel to the open break-out rooms for coding in the main session: A demo (45min) - Step-by-step demonstration of the procedure to build and deliver EBRAINS tools for the Lab: creation of Spack package, contribution to the EBRAINS official software release repository, automated test builds on gitlab runners, CI/CD workflow, validation in the lab-int and deployment to production and following Q&A.

(dial in)

Analysis of cortical wave activity using the CoBraWAP pipeline

Speakers: Giulia De Bonis, Michael Denker

The availability of neuronal data describing brain activity is higher today than ever before. This acts as an enabler of new approaches in neuroscience research. Given that data is levied on a multitude of different spatio-temporal scales, using a highly diverse methodological toolset on the experimental side, one of these approaches is to concentrate on the detailed comparison and integration of such heterogeneous activity data. Scientifically, this presents the challenge of relating the various recorded observables to one another in the context of the underlying model assumptions. On the technical side, the challenges we face are to strengthen rigor and reproducibility using standardized, high-performance pipelines. This ensures that results from different data modalities can be analyzed in a consistent manner, and thus meaningfully compared, while exchanging, mixing and comparing analysis approaches coming from the different neuroscience sub-domains. Taken together, aligning data and analysis from different sources in a reusable workflow, we can build a broad basis for meta-studies, contextualization of individual studies, and re-use the pipeline in the context of model validation and model calibration.

Here we present the Cortical Brain Wave Analysis Pipeline (CoBraWAP, https://github.com/INM-6/cobrawap) as a concrete implementation of such an analysis approach to analyze wave-like spatial activity patterns observed in activity measurements of the cortex, integrating diverse data sources such as ECoG, calcium imaging, local field potentials (LFPs), or spiking neuronal network simulations. We elaborate in detail on the pipeline design and its level of modularity, which is based on building blocks that provide implementations of various analysis methods and processing steps. The components are matched by their input-output relations and can be arranged to address different data and different scientific questions. Further, we show how the pipeline blocks are implemented using multiple community tools such as snakemake [1], Neo [2], and Elephant [3]. We further showcase and contrast our current efforts and plans to integrate the pipeline into EBRAINS services using multiple approaches, including deployment on HPC infrastructures, and usage of the emerging workflow system. Finally, we showcase examples how this pipeline enables science in use cases involving slow wave activity in ECoG, calcium imaging and simulation [4], and motor cortical beta waves in LFP.

• [1] Mölder F, Jablonski KP, Letcher B, Hall MB, Tomkins-Tinch CH, Sochat V, et al. Sustainable data analysis with Snakemake. F1000Research. 2021; available from: https://f1000research.com/articles/10-33
• [2] Garcia S, Guarino D, Jaillet F, Jennings T, Pröpper R, Rautenberg PL, et al. Neo: an object model for handling electrophysiology data in multiple formats. Frontiers in Neuroinformatics. 2014;8:10.
• [3] Denker M, Yegenoglu A, Grün S (2018) Collaborative HPC-enabled workflows on the HBP Collaboratory using the Elephant framework. Neuroinformatics 2018, P19. doi:10.12751/incf.ni2018.0019
• [4] Capone C, De Luca C, De Bonis G, Gutzen R, Bernava I, Pastorelli E, et al. Simulations Approaching Data: Cortical Slow Waves in Inferred Models of the Whole Hemisphere of Mouse. arXiv. 2022; arXiv:2104.07445.

In the parallel to the break-out rooms ongoing in the main session:

• Workflows: Dashboard - Hands on to interactive mock ups for gathering user stories / user requirements. (Presenters: Sofia Karvounari and Eleni Mathioulaki)

(dial in)

• One of the parallel sessions: TVB-Cloud Coding TVB-Multiscale (Jil Meier, Dionysios Perdikis, Valery Bragin)

(dial in)

• one of the parallel sessions: HDC / Health Data Cloud workflows (Patrik Bey, Marc Sacks)