adding new adsorbents

[ltalirz]

As of today, the AutocompleteInput does not allow for entering new values that aren't yet in the list.
Following our suggestion, this has been discussed at the bokeh level, and we have green light for adding support for this directly in the widget bokeh/bokeh#10295 (comment)

While this is a bit of work, it's probably better than an ugly workaround?
At the same time, the question is: if we want a new material, will people also need to enter a new materials identifier etc.?

[dwsideriusNIST]

I can add the function that generates the materials identifier to the repo. As we discussed, the "name" ---> ID hashing sequence is not intended to be reproducible, so it is OK if the digitizer's hashing sequence differs from one I use in house. The only restriction is that the format be the same: "NIST-MATDB-[MD5 hash digest]"

NOTE: I checked my code, the material identifier is generated with the MD5sum hash. We can salt the string if desired.

[ltalirz]

I can add the function that generates the materials identifier to the repo.

Thanks, Dan, in that case I will simply copy it and apply it here, once the AutocompleteInput works accordingly.

[ltalirz]

NOTE: I checked my code, the material identifier is generated with the MD5sum hash. We can salt the string if desired.

Ah. Ok!

[ltalirz]

Example of adsorbent representation

{
		"hashkey": "NIST-MATDB-SHA2hashdigest,
		"name": "New Material 2 name",
		"Formula": "Material 2 Formula",
		"Synonyms": [
			"Material 2 First Syn",
			"Material 2 Second Syn",
			"Material 2 Third Syn"
		],
		"External_Resources": [
			{
				"resource_id": "bd45316e8286e6ae4329da7a3448bcbd7fe45e91",
				"name": "COD",
				"values": [
					"1100118",
					"http://www.crystallography.net/cod/1100119.html"
				]
			}
		]
	}

[ltalirz]

Started working on this here: bokeh/bokeh#10476

[ltalirz]

Just to mention that the PR has been merged and bokeh 2.3.0.dev1 includes it, but now we're stuck on the panel side - let's see for how long holoviz/panel#1588

I've tried setting up an environment with both bokeh and panel development versions but failed to get it to work so far...

[ltalirz]

@danieleongari @dwsideriusNIST After fighting with the bokeh & panel development versions for hours, I've finally managed to get a version deployed on the server that allows specifying new adsorbents.

See http://digitizer.matscreen.com/ (using panel-dev branch of digitizer + this panel branch)

[ltalirz]

#48 adds the ability to enter adsorbents of unknown name.

Still to decide how to ask the user for additional information regarding the adsorbent.

[dwsideriusNIST]

There are a few options to consider:

1. With the PHP digitizer, when a new adsorbent name was input (exactly the same behavior as the panel-dev branch), I left "hashkey" empty (though we should go with null). Any isotherms with empty hashkeys were flagged for review, after which I either corrected the adsorbent metadata or generated new hashkeys and updated the isotherms.

Main point: I'm OK with a null hashkey as the postprocessing script will flag that isotherm for review.

2. For more experienced users, there could be a breakout menu where additional information (CSD refcode, synonyms, etc) is input. The hashkey generation could be an option as well.

-> Perhaps the best approach for now is to leave hashkey as null and we develop the breakout menu at a future date.

These options would be equally applicable to adsorbates/InChIKey. (though it would be great if there was a robust API for cross-referencing InChIKey/InChICode/IUPAC names.)