a data shifting bug in your data for MOF_data..

[usccolumbia]

I tried to run your experiments on the MOF data and it reports an error, which is caused by the headline of the bandgap file
in ./data/MOF_data/qmof_database/opt-bandgaps.csv
it shows the bandgap of
refcode,BG_PBE,CBM_PBE,VBM_PBE,Direct_PBE
ABACUF01_FSR,0.635073,1.244977,0.609904,False
ABAFUH_FSR,2.1930389999999997,3.759203,1.566164,False
ABALOF_FSR,1.134232,3.43044,2.296208,False
ABAVIJ_FSR,0.345448,1.09114,0.745692,False
ABAVOP_FSR,0.342645,1.308492,0.965847,True

However, in your generated targets.csv file:
ABACUF01_FSR,BG_PBE
ABAFUH_FSR,0.635073
ABALOF_FSR,2.1930389999999997
ABAVIJ_FSR,1.134232
ABAVOP_FSR,0.345448

all bandgaps values in targets.csv are wrong....! shifted by one value.

does this mean your result in the paper is wrong?

[usccolumbia]

just need to change the data/MOF_data/process.py file line
targets.append([str(refcodes[i]), properties[i+1,1]])

[vxfung]

looks like a bug with the script to get MOF data, and what you suggest looks right to me; thanks for pointing this out. these do not affect the results in the paper, which predate the scripts.

[usccolumbia]

it's not clear how to download the Alloy data. Can you add more detail?

…

________________________________ From: Victor Fung ***@***.***> Sent: Friday, June 25, 2021 10:01 AM To: vxfung/MatDeepLearn ***@***.***> Cc: Hu, Jianjun ***@***.***>; Author ***@***.***> Subject: Re: [vxfung/MatDeepLearn] a data shifting bug in your data for MOF_data.. (#1) looks like a bug with the script to get MOF data, and what you suggest looks right to me; thanks for pointing this out. these do not affect the results in the paper, which predate the scripts. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub<https://protect2.fireeye.com/v1/url?k=bd3e976a-e2a5af95-bd3ed9ab-86da9d9f5e25-f5e1afff9a3ac531&q=1&e=e2ff70cc-f22a-4115-bd37-fb58e59312a0&u=https%3A%2F%2Fgithub.com%2Fvxfung%2FMatDeepLearn%2Fissues%2F1%23issuecomment-868521637>, or unsubscribe<https://protect2.fireeye.com/v1/url?k=74624022-2bf978dd-74620ee3-86da9d9f5e25-600c6b553d77ccfc&q=1&e=e2ff70cc-f22a-4115-bd37-fb58e59312a0&u=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FADXPNM7ZKS53BM6SXUPWTBTTUSD4ZANCNFSM47JG5HXQ>.