adswa / mpi-datamanagement-ws Goto Github PK

@lnnrtwttkhn mentioned Mkdocs, which sounds ideal for creating notes and material and rendering it. Deploying via Github pages sounds easy. Continuous integration with Travis should be doable.
Could you set it up initially, @lnnrtwttkhn, and then we'll see to get it deployed and tested?

A rough outline of what slides from the initial, general section should cover. Edit this comment for additions, tick of things that are done. Just as a mental place holder.

General welcome and logistics

• Welcome to the workshop
  • Introduce speakers
  • Introduce ReproNim
  • Overview of BBB (where is the chat, how to unmute, what are breakout rooms + mention that one needs to rejoin the main room after leaving a breakout room, ...), rules for interaction (mention CoC, determine how and when participants can ask questions)
  • mention again that the room will be recorded
• Test the live polling system
  • Ice breakers (+ affiliation, career stage, reason for joining)
  • WS-relevant info (OS, experience/knowledge about core concepts or tools)
• Overview of the workshop
  • Quick overview of topics that are covered plus time estimate
  • Approximate slots for breaks
  • mention that everyone can drop out and rejoin as they please

Concepts, Motivation

• Definitions
  • Research Data Management
  • Version Control
  • Provenance
  • FAIR principles
  • (Computational) reproducibility
• Motivation
  • How does RDM make research easier?
  • more reproducible?
  • better?

[TBC]

it's still a bit empty there ...

Name: Adina Wagner, Lennart Wittkuhn - Two joint workshops at our respective home institutions in Germany (MPI Berlin, Juelich Research Center). TEAMWORK

Who: neuroscience grad students up to PI level (whoever is interested) from home institutions + surrounding institutions (for Berlin, e.g., Berlin Universities, for Juelich, e.g. affiliated universities Aachen, Düsseldorf, Köln)

How: full-day workshop (remotely or physically, depending on covid-19 situation)

What:
Berlin: Basic DataLad workflow focusing on data management (e.g., from data collection to paper manuscript)
Juelich: Less data management centric (there are regular data management courses held by Adina anyway), with elements from the data processing module and sneak-peaks into fmriprep, mriqc, BIDS, and other open software tools

• Github repository: https://github.com/adswa/mpi-datamanagement-ws/
• Homepage: https://adswa.github.io/mpi-datamanagement-ws/ideas/
• Notes: https://docs.google.com/document/d/1Lw4o5Q3OMjXgGOyFBUHP--_xv01KJtsY76eViMJJLhw/edit

(When): MPI Berlin: A full-day workshop is tentatively scheduled for fall, likely virtual. A date for the Juelich Research Centre workshop is not yet settled.

We should schedule a test run for:

• Video conferencing system (I've seen issues today on BigBlueButton, so we should test, and migrate if needed)
• Code execution tool (if @lnnrtwttkhn wants to use it, too)

I think it would be nice if participants create data during the workshop themselves, collectively add this data to a shared repository and then run analyses on it.

This was done in a previous workshop on the R package repro that I attended before at the MPI (see details below) and I really liked this, from a didactic perspective:

https://github.com/aaronpeikert/repro-COMP2PSYCH/tree/master/R

Hi @adswa, as we discussed during our meeting today, I opened this issue on the question of video conference tools and computational resources for the workshop.

• What are suitable video-conference solutions when using live-coding?
• Are there any computational resources (cluster environments etc.) that could easily be used by the workshop participants?

Just realized that this could also bee two separate issues. We can split it if you prefer :)

Important points for virtual meetings: https://twitter.com/alexwlchan/status/1262485550231162880
Generally cool: https://www.humanbrainmapping.org/i4a/pages/index.cfm?pageid=3846

When going virtual, we likely need to have participant register such that they can get an invitation with a zoom/jitsi/whatever link and a password.
I've created a Google doc for us to keep a list: https://docs.google.com/spreadsheets/d/16QTbtTF7UsQ2ZSY0vcZb8rfGdOxoTPw6xcaNtdTl9y8/edit#gid=0
@lnnrtwttkhn, you should have gotten an invitation to your institutional email, let me know if I should resend it.

I have played around with possible ways to include material into mkdocs pages. Personally, I script my slides in HTML using reveal.js and save them as PDF. Unfortunately, markdown does not support embedding PDFs. I have tried to embed a PDF using raw HTML, but github.io seems to be blocking this:

<object data="https://github.com/datalad-handbook/course/blob/master/talks/PDFs/OHBM.pdf" type="application/pdf" width="700px" height="700px">
    <embed src="https://github.com/datalad-handbook/course/blob/master/talks/PDFs/OHBM.pdf">
        <p>This browser does not support PDFs. Please download the PDF to view it: <a href="https://github.com/datalad-handbook/course/raw/master/talks/PDFs/OHBM.pdf">Download PDF</a>.</p>
    </embed>
</object>

Result:

It is possible to add individual slides' HTML plainly into the markdown documents. This comes with many disadvantages, though: Raw HTML is not super readable, unless we spent some time wrangling the site theme, the slides loose much formatting (and html would need to be adjusted). Also, this location would then be a second location to be maintained and checked for accuracy of content.

There is obviously the nevertheless-to-be-used option of adding links to slides into the markdown sections for download, but we may want to write a bit of a syllabus around it (that needs to be written for maximum stability) to not only have empty websites with a few links in bullet points in them.

Requires only a browser, can be embedded into slides 👍

Open questions that we are happy to receive feedback on:

Content questions:

• What are strategies to keep everyone on-board during hands-on coding sessions?
  • Benchmarks for no. of TA's per number of participants?
• What are simple open datasets or tasks that one could use to produce data during the course that could be put under data management control with DataLad that are not specific to a subfield of neuroimaging (i.e., not everyone is familiar with fMRI)?
• Is it feasible to go from collecting data to paper manuscript within one session?

Practical questions:

• What are good video conferencing tools that one can use for a hands-on workshop? (see #4)
  • Audience response tools
  • Pre and post conference feedback
  • Conversations about expectations
• How can one ensure to not waste too much time on setting up computers and software etc.?
• What are computational resources (HPC) that one can easily use without too much preparation?
• Any more ideas on how can we make the most use of the training materials we create?
  • Workshop recordings to Youtube
  • Website with materials can stay online forever, but shall we migrate it to some place where it is more easily findable? Can we link our materials with ReproNim resources?
  • Contribute some of our materials back into the repronim modules, especially the "reprucibility Basics" module

hey @adswa, I know that we already settled for a date but I wondered if it would be a good idea to offer multiple dates (e.g., three to five) and then let everyone indicate their preferred dates during the registration?

given that we are rather flexible with the exact day, this might allow us to maximize the number of potential participants? the question about the dates could easily be included in the registration. what do you think?

I'll be collecting relevant information about a Juelich event in this issue.
People in my institution are highly supportive of a workshop, and the tentative plan is autumn of this year virtual or spring of next year in person.

CC-BY-SA-3.0?

• invite people to complain about missing or false stuff
• invite people to contribute content or fixes
• outline how to create and serve documentation locally (maybe add a requirements.txt to install from), how contributors are expected to contribute

The needed information include:

• No. of Attendees
• Training Topic
• Git Repo
• Expected Course Type
• Target Course Delivery dates
• Other Links
• Audience
• Computational Infrastructure
• Outcomes

I won't put the link to the Google spreadsheet here, since I am not sure if it can be shared but I think it's fine (and actually useful!) if we add these information to the documentation