It seems that the value of VSA_EState* was not generated well across the entire training data.

[jsko-arontier]

Thanks for publishing the good work. I ran the program to use ROBIN for our study and noticed something strange.

When I run the analysis using the provided file (Mordred_Test_Compounds_3D.csv), I get the results as stated in the paper, but when I run the analysis by generating descriptors directly from the sdf file, I get different results.

When I analyzed the generated files, I found that the VSA_EState* values were significantly different, as shown below, and in the provided files (Mordred_Test_Compounds_3D.csv, Mordred_ROBIN_RNA_Binder_3D.csv), the VSA_EState1~7 values are mostly 0. If you generate them yourself, these values will be present.

Here is the program I used

• rdkit : 2022.9.5
• mordred : 1.2.0
• tensorflow : 2.3.1
• scikit-learn : 1.0.2
• numpy : 1.18.5
• scipy : 1.9.3

$ cat Mordred_files cat Mordred_Test_Compounds_3D.csv | cut -d ',' -f 1,1561,1562,1563,1564,1565,1566,1567

name,VSA_EState1,VSA_EState2,VSA_EState3,VSA_EState4,VSA_EState5,VSA_EState6,VSA_EState7
ADQ,0.0,0.0,0.0,0.0,0.0,0.0,0.0
HIV TAR compound 4,0.0,0.0,0.0,0.0,0.0,0.0,0.0
Ribocil-A,0.0,0.0,0.0,0.0,0.0,0.0,0.0
Tetracycline,0.0,0.0,0.0,0.0,0.0,0.0,0.0
Imatinib,0.0,0.0,0.0,0.0,0.0,0.0,0.0
Ibrutinib,0.0,0.0,0.0,0.0,7.188619484542558,0.0,0.0
Lovastatin,0.0,0.0,0.0,0.0,0.0,0.0,0.0
Nevirapine,0.0,0.0,0.0,0.0,0.0,0.0,0.0

$ cat Mordred_files cat Mordred_ROBIN_RNA_Binder_3D.csv | cut -d ',' -f 1,1561,1562,1563,1564,1565,1566,1567 | head -n 5

name,VSA_EState1,VSA_EState2,VSA_EState3,VSA_EState4,VSA_EState5,VSA_EState6,VSA_EState7
0054-0090,0.0,0.0,0.0,0.0,0.0,0.0,0.0
0096-0280,0.0,0.0,0.0,0.0,0.0,0.0,0.0
0109-0002,0.0,0.0,0.0,0.0,0.0,0.0,0.0
0109-0045,0.0,0.0,0.0,0.0,0.0,0.0,0.0

[jsko-arontier]

The file I created myself has the following non-zero values.

$ python -m mordred -3 Test_Compounds_3D.sdf -o Test_Compounds_3D.mordred.csv
$ cat Test_Compounds_3D.mordred.csv | cut -d ',' -f 1,1561,1562,1563,1564,1565,1566,1567 

name,VSA_EState1,VSA_EState2,VSA_EState3,VSA_EState4,VSA_EState5,VSA_EState6,VSA_EState7
ADQ,11.718160420815336,47.42857699012404,34.05060062906003,3.078153942379781,-0.33326002742329375,14.99573145325298,-1.9379634082088582
HIV TAR compound 4,38.19259310235959,17.498854087497666,3.1033958228353073,6.24855622042333,-0.5625207860922143,7.987202028487653,-4.473388303949271
Ribocil-A,0.0,31.842610646889042,4.933083613714164,1.7610598832630966,1.6422973544097212,5.556640742998518,5.828923148026683
Tetracycline,0.0,40.01794688016008,54.16668723629131,-0.5148717473036521,-7.799810531338082,3.068299635644739,-0.19874055177626593
Imatinib,0.0,30.128187865789258,6.2979499406137265,6.102768934885226,0.3376392406237758,19.347018038601632,5.219265870655177
Ibrutinib,7.793632778010275,22.652446521864317,5.616525798335863,8.52059370897009,1.8032790912538088,17.321361934766436,4.581419830522775
Lovastatin,11.505992733717303,24.21316292324496,9.89343956573144,1.2726621945074328,0.7548182808576581,0.0,9.701063477760322
Nevirapine,0.0,15.195298563869994,12.347919501133788,2.1814871504157223,1.1875000000000004,5.805969387755102,5.664973544973545