Teaching microbial electrochemistry on real data, together

What we would like to agree

How this works

What a student does, in order. Each step shows whether it exists today, taken from the inventory and the modules on this page. A step that is only planned has no link.

1. 1

   Start at the bench

   Partly built

   Follow the first experiment a student runs at the bench: one single MFC, from its materials to its first power curve and coulombic efficiency. The chart below lays out its steps in order, from materials to measurement.

   • Status from module (a), “From bench to simulator”: Part live.

   Open the chart
2. 2

   Read a step

   Partly built

   Select a step in the chart. The side panel gives its measurements and, where the graph records them, its equipment, precautions and references (a reference the corpus holds links to its record); the line under the chart says what a student learns there.

   • Status from module (a), “From bench to simulator”: Part live.

   The papers behind the chart
3. 3

   Model it in the 3D lab

   Available

   In the 3D lab, choose a model and change its electrodes, chamber, membrane, organism and circuit; the prediction updates as you change them. The four models further down show the devices in the order we would teach them.

   • Status from “3D lab” in “What exists today”: Live.
   • Three ready-made templates, one for brewery effluent; the UFZ setup is not one of them yet.

   Open the 3D lab
4. 4

   Check an extracted value against its paper

   Planned

   Take a value MESSAI extracted from a paper, read the sentence it came from, and decide whether it is right, wrong, or cannot be told from the quote.

   • Status is that of the least built part, module (b), “Verify a paper”: To build.
   • Also relies on “Literature corpus” in “What exists today”: Live.
   • Today only an admin review queue writes a verdict, and it keeps no reviewer.
5. 5

   Score how a paper reports its method

   Partly built

   Paste a paper’s DOI into the reproducibility scorer. It scores the paper against the five-parameter ISMET minimum, an 18-parameter view and a model per kind of study, and marks methods and equipment for a manual check, which is the part a student does.

   • Status is that of the least built part, module (c), “Score a paper against its reporting checklist”: Part live.
   • Also relies on “Reproducibility scorer” in “What exists today”: Live.

   Open the reproducibility scorer
6. 6

   Tell us what is wrong

   Available

   Say what is wrong or confusing. The feedback widget on most pages records the page, the kind of feedback and your comment (it is not yet in the lab or the learn section), and signed-in users can also comment at the foot of this page.

   • Status from “Feedback widget” in “What exists today”: Live.

   Comments on this page

What exists today

Every row was checked against the repository on 8 Oct 2026. “Live” means a reader can use it now, “In progress” that it exists but is not finished, “Planned” that only the roadmap names it. Most of this was built for research and engineering, not as a course; the modules below turn it into one.

Proposed modules

Five modules we could build together, each small enough to test on one cohort. Where the platform already does part of the job, the module says so. Effort is a rough estimate for one person part-time; the UFZ side is time to specify, review and supervise.

1. (a)

   From bench to simulator

   Part live

   Learning goal

   Predict before you measure. Students rebuild a real UFZ setup in the lab, state the model’s prediction with its interval, then compare it with the measured run and explain the gap.

   Format

   One two-hour practical in pairs, with a worksheet and one measured run supplied by UFZ.

   Builds on

   The lab and its templates, “Save as Experiment” and the run page, where measured results sit beside the stored prediction (sign-in). The model gives no number when an input is missing, which is itself a lesson. Most published presets are calibration cases, so a UFZ run the model has not seen is a fair test.

   MESSAI builds

   The UFZ setup as a lab template and the worksheet. With the measurement upload (roadmap C3-M1, 21 Nov 2026) and the measured-versus-predicted view (C3-M2, 19 Dec 2026), a comparison over a whole run.

   UFZ contributes

   The setup (geometry, electrodes, medium, operating conditions) and one measured run with its raw data, as the group would hand it to a new student.

   Effort

   MESSAI about one week; UFZ about half a day to specify the setup.

2. (b)

   Verify a paper

   To build

   Learning goal

   Read critically. Students check values MESSAI extracted from a paper against its text and decide for each whether it is right, wrong, or cannot be told from the quote.

   Format

   A seminar hour plus homework: three to five values each from a paper the supervisor chooses, and a one-page report of what they found.

   Builds on

   The nitrogen page is the case. Of 1,025 modelable nitrogen facts across 490 papers, none is marked verified (staging snapshot of 5 Oct 2026), and 76 come from a legacy import never grounded against full text. Only 164 of the 490 papers have a PDF stored. The onboarding overview’s research route already ends in a first task of this kind: trace one measurement back to the sentence it came from. The nitrogen page’s worked paper shows what students find: a removal figure whose quote carries no number, and a stated power and current density that imply about 375 V from one cell.

   MESSAI builds

   The paper selection, a checking guide with the failure types to look for, and a way to record a student’s check. Today only an admin review queue writes a verdict, and it keeps no reviewer, so the recording has to be designed with UFZ.

   UFZ contributes

   A senior reader to confirm what students flag, and the papers: start with the group’s own topics.

   Effort

   MESSAI about one week, including the recording; UFZ about two hours to choose papers, then supervision.

3. (c)

   Score a paper against its reporting checklist

   Part live

   Learning goal

   Know what a reproducible methods section contains. Students score a paper against the checklist for its kind of study and name what they would need to repeat the work.

   Format

   A seminar exercise, about 30 minutes per paper: score by hand first, then compare with the automated score.

   Builds on

   The reproducibility scorer: the five-parameter minimum, an 18-parameter view and a model per kind of study (14 kinds, scored on conditions, outcomes, methods and equipment). Methods and equipment live in the text, so the automated score marks them for a manual check. That is the part students do.

   MESSAI builds

   A scoring sheet with an answer key for three to five papers, and a view of where the cohort’s scores differ from the automated one. It feeds the reporting checklist v1 (roadmap B3-M1, 10 Oct 2026).

   UFZ contributes

   The papers, including one students will meet in their thesis, and a judgement on which checklist items matter for electrobiotechnology.

   Effort

   MESSAI about three days; UFZ about two hours to review the items.

4. (d)

   Potentials and units primer

   Part live

   Learning goal

   Avoid two comparison traps: values on different reference electrodes, and rates normalised to different areas or volumes. Students convert, then say whether two numbers can be compared at all.

   Format

   A 45-minute primer and an exercise sheet of paired values taken from real papers, with worked answers.

   Builds on

   The onboarding overview already ends its science route with this task: read one performance figure and say which normalisation basis it uses. The onboarding atlas covers the redox ladder on the SHE scale and “the normalisation trap”. The data model stores the reference electrode and the normalisation basis with a value, because they decide whether two voltages or two power densities compare. How often each error occurs in the corpus has not been counted.

   MESSAI builds

   The exercise sheet and answers, built from corpus rows where the reference electrode and the basis are recorded, in German and English.

   UFZ contributes

   The reference electrodes and conventions the group’s students actually meet, and the mistakes you have seen them make.

   Effort

   MESSAI about three days; UFZ about two hours to review.

5. (e)

   Thesis onboarding path

   To build

   Learning goal

   A new PhD student reaches a defined starting point in the first weeks: knows the field’s vocabulary, the group’s setups and the platform’s tools, in German or English.

   Format

   A self-paced path of five stations (read, see, simulate, check, report) with a short self-check at each, ending in one hour with a supervisor.

   Builds on

   The learn hub, the onboarding overview, the Systems Atlas, the 3D lab and its templates and the microbe catalog, with modules (a) to (d) as stations. The ISMET member path (roadmap D1-M4, 19 Dec 2026) keeps the persona and marks sign-ups from ISMET.

   MESSAI builds

   The path as one page in both languages, built from the stations, with the self-checks.

   UFZ contributes

   The group’s own list of what a new student must know in the first month, its setups and protocols, and a German-language review.

   Effort

   MESSAI about one week once modules (a) to (d) exist; UFZ about half a day for the list and the review.

A first MFC, step by step

The backbone of module (a), From bench to simulator. This is the existing /protocols graph, cut down to the first experiment a student runs at the bench: one single MFC, from its materials to its first power curve and coulombic efficiency. The step names and descriptions are the graph’s own English text; the sentence beside each step, on what a student learns there, is written for this page. Columns run from materials to measurement.

Models

Four models, in the order we would teach them. Each changes one thing from the one before. Drag to orbit; hover a label marked ⓘ for what the part does. The geometry is illustrative, not a particular lab’s reactor.

Papers behind this page

Every paper the protocol chart cites, with what it is cited for and where you can check it. Each row links the publisher and, if this site holds the paper, its corpus record.

Our to-dos (MESSAI)

In priority order. Estimates are rough and assume one person part-time; those with a roadmap id are the roadmap’s own.

1. Agree this proposal~1 dayThis week

   Turn this page into a one-page agreement of who contributes and gets what. The ISMET partnership proposal (roadmap D1-M1) is due 9 Oct 2026. Everything else waits for the agreement.

2. Education module v1~3 days (roadmap D1-M3)Planned, due 28 Nov 2026

   One guided tutorial on the resource-recovery demo and the lab, with a worked brewery exercise and a worksheet. It sets the format every module reuses.

3. The UFZ setup as a lab template~2–3 days after the specWaiting on UFZ

   Geometry, electrodes, medium and operating conditions from the group, encoded so that students can load them. Parameters the lab does not expose are listed, not guessed.

4. A way to record a student’s check~1 week plus a migrationNeeds a decision

   Decide with UFZ what “checked by a student” means in the data (verdict, reviewer, date, who confirms), then add it. The schema change goes to staging first.

5. Exercise sheets and answer keys~3 days eachNot started

   Modules (c) and (d), built from real corpus rows, in German and English. Each answer cites the paper it comes from.

6. Measurement upload~1 week (roadmap C3-M1)Planned, due 21 Nov 2026

   Upload a CSV or JSON of a measured run so that it is stored as “measured” and shown beside the prediction.

7. Feedback widget in the lab~2 days (roadmap A1-M4)Planned, due 31 Oct 2026

   Mount the widget in the lab and add the usage page, so that lab exercises produce a feedback signal and the pilot has a readout.

8. The thesis onboarding path~1 weekLater

   One page in English and German assembling the stations, after modules (a) to (d) have run once. The ISMET member path (roadmap D1-M4, ~2 days) reuses it.

Deliverables

Dates from the roadmap are fixed. The rest count from P0, the first day of a pilot cohort we agree together, so nothing is promised before a cohort exists.

1. This weekProposal for review

   This page, the module outline and the one-page agreement of who contributes and gets what.

2. 28 Nov 2026Education module v1 (roadmap D1-M3)

   A guided tutorial on the resource-recovery demo and the lab with a worked brewery exercise; feedback captured through the widget.

3. 19 Dec 2026ISMET member onboarding path (roadmap D1-M4)

   The persona is kept and sign-ups from ISMET are marked as such.

4. P0 − 4 weeksUFZ setup and sheets for review

   The setup as a lab template and the exercise sheets for modules (a) and (c), for UFZ to correct.

5. P0 − 2 weeksDry run

   One or two group members work through the modules; the German and English text is frozen.

6. P0Pilot cohort starts

   Modules (a) to (d) run; the feedback widget is on and we log what we can.

7. P0 + 4 weeksReadout

   What confused students, what they got wrong, what we change, with the measures below.

8. P0 + 8 weeksRevised modules and the thesis path

   Revised modules and module (e) for the next cohort. Offered to ISMET members (roadmap D1-M5, phase P1) if UFZ agrees.

What we ask of UFZ

• A lecturer or senior researcher to co-own the pilot, who sets the learning goals, reviews every exercise and decides what is worth teaching.
• A pilot cohort: a seminar, a lab rotation or a group of new PhD students, with a start date and a size we agree together.
• One or two lab setups to model, each with geometry, electrodes, medium and operating conditions, and for module (a) one measured run with its raw data.
• Three to five papers for modules (b) and (c), including at least one that students will meet in their thesis.
• Feedback from the students’ side: where they got stuck, what they skipped, and a review of the German text.
• Agreement on credit and data: who is named on the modules, whether student checks may enter the corpus, and what we record about individual students. We propose aggregate counts only.

How we would know it worked

Rows marked Roadmap are the targets in MESSAI’s roadmap (workstream D1). Rows marked Proposed are ours and have no target yet. Setting one before a first run would be invention, so we propose to set them with UFZ after the pilot.

Open questions for UFZ

1. Does the group already teach a course, a lab rotation or an onboarding for new PhD students? What would this replace or extend?
2. Which of the five modules would you use first, and which would you drop?
3. Which setups should students rebuild, and may their geometry and raw data appear on a public page, or only behind sign-in?
4. What should “verified by a student” mean in the data? Would you accept a student’s check as a first screen that a senior reader confirms before a value counts as verified?
5. Which reporting items matter most for your methods, Kolbe electrolysis and electrosynthesis among them, and which of MESSAI’s checklist items would you change?
6. Language: German first for the thesis path, or both from the start? Who could review the German text?
7. Licence and reach: should the modules be open (MESSAI’s learning package is CC BY 4.0) or for ISMET members only, and which other groups should join the pilot?
8. Timing: when does the next cohort start, and is there a window before 28 Nov 2026?