AN OPEN RESEARCH INITIATIVE
Questions to investigate. Knowledge to share.
A space taking shape to bring together questions, evidence and contributions from different people. We are starting with spinal cord injury and repair; other research lines will follow as work, participation and resources develop.
Explore the first research lineOne step at a time
Each research line starts with a specific question. We review sources, check analyses and document both progress and unanswered questions. We want to grow through contributions that can be reviewed and used in the next step.
First research line · Spinal cord injury and repair
Our current work is documentary and computational. Within this first research line, we have developed two areas, each with results and limitations that can be explained.
- Materials and transport
- Models of how signals are distributed and transported within a material. E001 and E002 are complete; calibration still requires measurements from the same system.
- Cell states and regeneration
- Reviewing expression data and their links to animals, samples and anatomical observations. Identifying a cell or an injury response does not demonstrate functional recovery.
We will add other research lines when they have a question and concrete work. Their topics and fields will be organised as participation grows.
Progress and open questions
Updated
From literature review and reproduction of a published analysis, we have moved to our own models and checks of cellular data. Each result is kept with its sources and limitations.
Quantum biology: reproduced a planarian model and tested its sensitivity to physical assumptions. The calculation agrees with the published results; the biological mechanism remains unidentified.
Data provenance: technical links verified for 330 Kim columns and eight Squair libraries. The missing biological keys remain unresolved; the records needed next have been documented.
Cell states: five published counts reproduced. Gap43 analyses do not support a stable association under the checks examined; no regeneration predictor has been validated.
Materials: relevant public methodological resources checked again. Calibration gaps remain; the pilot is still documented, not performed.
Quantum exploration · Results and limits
Exploration · Quantum biology
We are exploring whether molecular spin processes could influence signals related to regeneration. This is an exploratory question; a connection to human spinal cord repair has not been demonstrated.
What we checked
We reproduced 947 points across three published curves from a radical-pair model in planarians, plus one case from its supplement. Changing the nuclei and spin relaxation substantially changed the magnitude and, at some fields, the direction of the effect.
What is missing
The sources reviewed do not experimentally identify the radical pair or measure its spin relaxation rates in the planarians studied. Detecting superoxide does not by itself identify the mechanism producing it. A simple empirical alternative also reproduces the pattern of increases and decreases, although it fails in magnitude.
We retain the hypothesis and the reproducible work. Further work on this model awaits molecular evidence or a prediction that can distinguish explanations. This is not a therapeutic result.
Cell states · Results and limitations
What could be checked
Five counts from Squair 2023 were reproduced exactly using the deposited annotations. This is a partial reproduction: it does not rerun sequence processing or validate anatomical or functional results.
A reanalysis with six mice did not support a stable Gap43 signal in the chosen contrast. In a separate dataset, Kim, 326 experimental cells were verified with labels and 28 animal codes; the other four columns are references. The contrast using 20 animals was sensitive to quality criteria. These are separate analyses, not predictive validation.
What the provenance search resolved
We verified 514 sequencing records: 495 from Kim and 19 from Squair. Kim's histology inventory enabled review of 15 tabular files with matching hashes. Neither these files nor the public code history resolved the mapping from original names to GEO, batches and final exclusions.
In Squair, the 12,447 metadata rows are assigned to eight libraries, but 4,727 use repeated NA identifiers. They were not removed or interpreted as biological duplicates. A manifest linking animals, pools, preparations and anatomy is still missing; discrepancies between sources also remain.
The next records needed
An explicit mapping and exclusion table for Kim, and the origin of each Squair library. Without those keys, expression analysis is not repeated or attributed to independent animals. We have documented the limit, not shown that those records do not exist.
Materials · Models, measurements and pilot
Evidence reviewed. Better measurements come next.
Updated
- Completed
- E001/E002 models and a review of published data, compatibility across experiments, and physical mechanisms. Calculations and their reproducibility have been checked.
- What is missing
- Linking the actual dose, hydrated geometry, storage capacity and solute mass balance in the same samples. The recovered data do not yet support a joint physical calibration.
- Next step
- Additional experimental records have been requested. If they do not resolve the gaps, assess the feasibility of an acellular pilot that measures storage and transport separately.
The evidence combines published data and idealised simulations. The pilot is documented, not performed. Axon guidance, functional recovery and clinical efficacy have not been demonstrated.
Recovered data · What changes in the research
We reviewed hydrogel studies and repositories to connect diffusion, storage and supply for the same solute. We recovered and verified 33 mass records and 12 swelling-volume records from a PEG-VS study. These measurements were published by another team; they are not our own experiments.
These records show why hydrated geometry matters: a gel's dimensions can differ from its initial dimensions. However, the water a gel absorbs does not tell us how much solute it stores, and the swelling specimens could not be linked to each transport insert.
What the other disciplines contributed
Metrology and photophysics. Fluorescence intensity depends on the medium and illumination as well as tracer quantity. Calibration and mass recovery must be checked before interpreting a difference as transport.
Polymers, mechanics and fluid dynamics. Size selection can change a mixture's apparent mobility. Swelling changes geometry; liquid mixing and the membrane can limit access to the gel. None of these explanations was established as the cause of the historical discrepancies.
Quantum physics. Its concrete contribution concerns the fluorescent indicator's electronic response and how to represent it in the measurement. We found no evidence of an additional quantum transport mechanism in these data. This is not a spinal cord repair result.
Inverse problems. Different combinations of storage, transport and access can produce similar curves. Simulations support varying thickness and mixing, and measuring storage independently. This separation is a prediction for an experiment, not an experimental result.
We reproduced 23 analysis outputs: 22 were byte-for-byte identical, and one summary matched except for computation time. Reproduction checks the calculation; it does not by itself validate mechanisms or materials.
A pilot designed to distinguish the explanations
The design uses PEG-VS and 20 kDa fluorescent dextran as a methodological reference. It proposes measuring how much solute the gel retains in a small bath and recovering it at the end; then comparing two hydrated thicknesses and two mixing conditions, measuring both chambers and the full mass balance.
A release experiment using independent samples would be held out to test the prediction without refitting. Measurement sensitivity must be checked first. The experiment has not been performed or commissioned, and this reference does not amount to selecting a material for spinal cord repair.
E002 · Connections alone are not enough
We completed all 632 planned scenarios. We found small effects that depend on supply and design. No improvement exceeded the predefined margin against matched cavities. Reporting this result also helps direct the next work.
For readers who want the method and its limits
We compared ordered connections and eight random layouts with blind cavities—open to their own channel, but not joining the two channels—and with widened channels. The outcome was the time-averaged solute exposure of the less supplied of two fixed bands.
632 converged scenarios: 480 primary cases, 80 physical sensitivities and 72 position changes. We ran 1,896 integrations and audited 1,942 numerical files, including control fields. Mass balance, positivity, and spatial and temporal convergence were checked.
Of 568 comparisons, two exceeded the improvement margin against widened channels, only with imposed dose and unequal supply; none exceeded it against matched cavities. The other 566 remained below the margin. This does not mean all differences were zero.
The margin combines estimated numerical error with a floor of 0.01 in normalised exposure. It is a study rule, not a biological threshold. The model is two-dimensional and dimensionless, without experimental calibration. Improving average exposure did not improve every local indicator.
The eight random layouts are not biological replicates. We make no novelty claim. Technical documentation and outputs are preserved locally; a public download is not yet available.
E001 · Initial preparation matters
Across 342 scenarios, redistributing binding sites changed free protein. Initial preparation and where the outcome is measured can change which design is favoured: there was no general winner.
This model distinguishes free and bound protein. It does not reproduce release caused by fibrin degradation or a calibrated experimental material.
Brief research log
Multidisciplinary review completed: additional swelling evidence, signal controls and a pilot designed to distinguish competing explanations. Further records requested; experimental calibration remains pending.
Published data reconciled: nominal recipe correspondence and duplicate records resolved. Preparation, geometry and mass-balance gaps remain.
E002 completed: 632/632 scenarios converged. The predefined criteria do not support a general connectivity advantage.
E001 completed: spatial effects depend on initial preparation. Measurements are needed before optimising a formulation.
Defined the materials and AI programme and its first controls.
Reproduced a published figure as an exercise in verifiable analysis.
Participate and support
We want to build a community around specific questions. Contributions can include reviewing a source, reproducing an analysis or helping explain a result. The first step is for each contribution to have a clear scope and be open to review.
- Propose and collaborate
- Suggest a research direction or project, share evidence and data, or help review methods and results. Proposals will be assessed for their contribution and feasibility.
- Contribute capabilities
- Knowledge, data analysis, time and available tools, including AI tools when useful for a specific task.
- Provide financial support
- We will seek funds for specific tasks. Each request will explain the work, its cost and how the outcome will be reported.
Contributions are received through the PayPal account of Juan Manuel Caro González, initiator of VERA for Science.
Choose your amount on PayPal’s secure donation page. You can donate with PayPal or choose a credit or debit card without signing in to PayPal, when that option is available. The link opens in a new tab.
A contact channel is still to be set up.
How it began
This project grew out of living with a spinal cord injury and working with AI every day. The aim is to contribute to a shared question and honestly document what we learn.
Read the personal story
I was injured at 20. I lost my youth. My family has always stood by me; their presence is an essential part of this story.
I work in data analysis and digital transformation. I use artificial intelligence every day in my work. That experience makes me ask what I could contribute.
This is my last chance. That is how I feel about this opportunity to participate, learn and look for answers. So I gave AI a goal: regain all my functions before I turn 60.
I want to fulfil that dream and help others living with the same condition achieve it too. I do not know whether it will be possible, or when. I share it as my personal goal.
I have no idea where all this will lead me. But I am convinced that by bringing people together and using AI rigorously, we can help accelerate research.
Sources for the work presented
References for the first research line, on spinal cord injury and repair. Our own models are idealisations; they do not reproduce these experiments. Sources for the exploratory quantum-biology work in planarians are also included; that work does not demonstrate spinal cord repair.
- Rishabh et al. · PNAS Nexus 2026
- Zenodo · Planarian model code
- Usselman et al. · Scientific Reports 2016
- Player & Hore · Journal of Chemical Physics 2019
- Kattnig · Journal of Physical Chemistry B 2017
- Squair et al. · Science 2023
- GEO · GSE198949
- Kim et al. · Neuron 2023
- GEO · GSE205769
- PEG-VS study · Hydrogel transport and properties
- Figshare · Reviewed mechanics and swelling data
- NIST · Quantifying fluorescence intensity
- Ross et al. · Microporous hydrogel scaffolds
- Hui et al. · Spatial control of mechanical properties
- Johnson et al. · Signal retention and release