Project discovery: suggestions from biologist and capsuleer

[Max_Adoudel]

Hello, CCP.

Thank you for this new iteration of Project Discovery. I loved exoplanets, loved proteins (and actually sometimes use the database we created back then) but now we once again on the biological side.

I am an immunologist. Not related to COVID, but nevertheless I know what cytometer is and how to analyze data it generates. And right now I have 4 improvement I want to suggest.

First of all - make more types of test samples. You feed us with all kinds of stainings (some pictures look familiar, but I won`t bet I know what is stained), which generate 1 cloud, 2, 3, 4 and so on. But you analyzed yourself just one easily distinquishable type. Those who just want profit from PD will rapidly learn that and the other samples from those high-rated ‘explorers’ will have only one small gate in the middle of nowhere. Take your time, make “test analyses” in all the clusters of images you have.

Secondly - give us more info. As was with star light curves, you gave in the corner possible types of stars. Here - give us info for stains - what was colored and why. It will improve understanding. But do this in the corner, so it won`t bother those who do not need this.

Third thing - give me more control while gating. I want to have ability to delete vertices by right-clicking on them while I do the gate. I want to auto-close the gate by right-clicking in space.

And lasty - give me more info while gating. I want to see intersections while I move the vertice, not when I finish. I want to see how the gate will look like if I close it now (faded line between pointer and first vertice will do).

Thank you. My suggestions on exoplanets on PD panel on FanFest 2018 were not taken to consideration. I hope you will hear me now.

[Lex_Paktu]

BTW you can cancel the cluster you are currently marking by clicking anywhere outside the window with the data, be it the panel on the right side or outside the project discovery window altogether.

Some of these suggestions will require further scientific work on their part, and these might take time or not be done. Though of course they would be nice.
On the other hand, I fully support all suggestions related to UI and control/info when gating.

[Max_Adoudel]

That cancelling annoy me actually) I can’t see borders of work area and when I try to work fast I sometimes erase the gate.

Scientifically it is not relevant if the gate catches big empty space and here you can go ‘too big’ and fail to create the gate.

[Dragomir_Krividus]

@Max_Adoudel

First of all - make more types of test samples. You feed us with all kinds of stainings (some pictures look familiar, but I won`t bet I know what is stained), which generate 1 cloud, 2, 3, 4 and so on. But you analyzed yourself just one easily distinquishable type. Those who just want profit from PD will rapidly learn that and the other samples from those high-rated ‘explorers’ will have only one small gate in the middle of nowhere. Take your time, make “test analyses” in all the clusters of images you have.

This, I agree on most other points too but this one is particularly annoying.
Give us more examples on how to divide more complicated samples. Also some examples are contradicting or misleading, sometimes the dividing line is placed differently on two very similar samples.

Edit:
From the other thread (New Project Discovery - golden zones - #7 by Leandra_Stormbreaker) it looks some characters get only limited set of examples to train on. Someone @CCP should look at that.