I agree in priniciple, but I am very scared of the natural laziness of mankind that will eventually lead to wrong usage (this is a general huge issue with the current trend of LLM and generative AI ..my AI-xiety is almost topping up my ecoanxiety right now...)

you nail it...exactly my impression (and fear)

You may claim that it is just supposed to assist human decision. But if you want to bring a human input, you will anway have to deep dive in the paper, read it and check the data/results. So...going to my initial claim, will it really help us ?... I am not sure...

If the standard will become that every editor/reader will first "qed" the papers, we'll hand up in some absurd system where the editorial decision and judged quality of a paper will mostly lies in details and how much everyone prepolish the article following black box statements.

Basically, it is not really asking : based on all the observable, which alternative models could be proposed to explain all these data ? (which is the key question when you review an article).

My global impression that that it tends to "comparmentalise" the studies overfocusing on every small statement/observation on their own, while not interogating the global working model with all the experiments/observable as a whole (which admitedly some human reviewers also do)

There are many essential things that we do as reviewer which is not provided here : evaluating the novelty and positioning relative to the field, ponderating the request for experiments with their difficulty and feasibility, and above all judge the statement with the full picture

So the wing disc overgrowth phenotype after all may just a be an ECM phenotype 😱😱😱 (ok there is a bit of a stretch here...). Beautiful story !

Big thanks for the support of our institution @pasteur.fr @cnrsbiologie.bsky.social and funding agency @erc.europa.eu and @agencerecherche.bsky.social who made this work possible.

Overall, our main message is that there are different flavours of mechanical competition where either growth or interfacial tension can contribute to WT cell compaction and elimination with distinctive geometrical and cell death distribution features.

Checking more carefully MyoII and tension, we found that N-cad induction also triggers a downregulation of MyoII and tension in junctions "orthogonal" to the clone boundary (for reasons we don't quite understand yet). Putting this feature in the vertex model fully recapitulate the pattern of death

There was one little feature though that was not recapitulated in our model: cells in the first row near N-cad clone interface tend to be bigger and are not eliminated. How to explain this ?

Alternatively, ectopic expression of N-cad was sufficient to increase boundary tension and could trigger WT or N-cad cell elimination as long as they are located in concave boundary/pockets. Basically, who is eliminated is fully set by the geometry of the interface (if you are in minority you lose)

Can we now find experimental conditions reproducing "pure" growth-driven or interfacial tension driven elimination ? First, activation of Yki/Yap in clones was sufficient to recapitulate all the features of growth driven compaction (no impact of curvature, elongation of cells parallel to long axis)

While both conditions could lead to cell compaction and elimination, clear distinctive features emerge in term of distribution of cell death, relationship with boundary shape and orientation of WT cell in the patch (check the paper to get the list)

So...we have now two contributors to WT cell compaction : growth and interfacial tension. Could we extract distinctive features from these two modes of cell deformation ? Alexis started to explore conditions in vertex model either increasing interfacial tension (below) or by increasing growth.

In this framework, just like the interface between two fluids, the pressure should be proportional to the local curvature. This is exactly what Léo found: WT cells near concave interfaces are more compacted and more likely to be eliminated

By imaging conditions with large clone coverage, Léo realised that WT patches surrounded by Ras cells tend to round up and sort. This is driven by the increase of tension at the interface of the clone. Could this generate a local increase of pressure (similar to Laplace pressure) ?

To sum up in few words, previously we found that activation of Ras in clones leads to neighbouring cell compaction and elimination. While we assumed that growth was the main driver of compaction, it was not clear whether other features could contribute to cell elimination