ics admet prediction tdc

[innoplexus]

@amva13 This PR is for submitting new entries/ new model to the TDC ADMET Group Leaderboard.

Please find the details below as per guidelines provided here

A. A link to the codes for training the model and running the benchmarks.
https://github.com/Innoplexus-opensource/ics-admet-prediction-tdc/tree/main

B. Instructions for replicating your results running these codes.
https://github.com/Innoplexus-opensource/ics-admet-prediction-tdc/blob/main/README.md

C. Description of the hardware used for training the model.
Machine with Intel i9-13900 processor 62GB RAM and NVIDIA RTX A6000 GPU with 48GB VRAM was used for the model training experiments.

D. Link to paper to be included on the website
https://github.com/Innoplexus-opensource/ics-admet-prediction-tdc/tree/main/report/Innoplexus ADME-NN Model Evaluation on TDC Benchmark ADMET Group Binary Datasets.pdf

E. Paper authors and developers of the model.
Rohit Yadav, Divya Nair, Amit Agarwal
Innoplexus Consulting Services Pvt Ltd, a Partex company

F. The dictionary string output of running the TDC benchmark group evaluations.
"Innoplexus ABDNN-ADME" (New model):
{{'caco2_wang': [0.148, 0.007]}, {'bioavailability_ma': [0.955, 0.017]}, {'lipophilicity_astrazeneca': [0.227, 0.018]}, {'solubility_aqsoldb': [0.387, 0.035]}, {'hia_hou': [0.985, 0.029]}, {'pgp_broccatelli': [0.985, 0.006]}, {'bbb_martins': [0.978, 0.004]}, {'ppbr_az': [3.744, 0.152]}, {'vdss_lombardo': [0.715, 0.043]}, {'cyp2c9_veith': [0.958, 0.006]}, {'cyp2d6_veith': [0.936, 0.013]}, {'cyp3a4_veith': [0.975, 0.005]}, {'cyp2c9_substrate_carbonmangels': [0.913, 0.038]}, {'cyp2d6_substrate_carbonmangels': [0.983, 0.013]}, {'cyp3a4_substrate_carbonmangels': [0.975, 0.008]}, {'half_life_obach': [0.445, 0.078]}, {'clearance_hepatocyte_az': [0.789, 0.019]}, {'clearance_microsome_az': [0.783, 0.039]}, {'ld50_zhu': [0.21, 0.017]}, {'herg': [0.963, 0.013]}, {'ames': [0.983, 0.004]}, {'dili': [0.998, 0.002]}, {'herg': [0.966, 0.009]}}

Kindly review and merge this PR. Also earlier we submitted "Innoplexus ADME" (Old model), There were some benchmark rankings missing from the leaderboard, i will provide additional PR for that as well for correction.

@amva13 the below PR is for correcting old entries/ old model to the TDC ADMET Group Leaderboard.

A. A link to the codes for training the model and running the benchmarks.
https://github.com/Innoplexus-opensource/ics-therapeutics-data-commons/tree/main

B. Instructions for replicating your results running these codes.
https://github.com/Innoplexus-opensource/ics-therapeutics-data-commons/blob/main/README.md

C. Description of the hardware used for training the model.
Machine with Intel i9-13900 processor 62GB RAM and NVIDIA RTX A6000 GPU with 48GB VRAM was used for the model training experiments.

D. Link to paper to be included on the website
https://github.com/Innoplexus-opensource/ics-therapeutics-data-commons/blob/main/Innoplexus_ADME_Model_report.pdf

E. Paper authors and developers of the model.
Rohit Yadav^1, Divya Nair^1, Prajakta Mate^1, Rushikesh Chaudhari^1, Arpan Sheetal^1, Sudhir Mansukh^1, Amit Agarwal^1, Anand Muglikar^2
1.Innoplexus Consulting Services Pvt Ltd, a Partex company
2.Perpetualblock Technologies Private Limited, a Partex company

F. The dictionary string output of running the TDC benchmark group evaluations.
"Innoplexus ADME" (Old model):
{{'caco2_wang': [0.289, 0.005]}, {'lipophilicity_astrazeneca': [0.499, 0.003]}, {'solubility_aqsoldb': [0.771, 0.005]}, {'ppbr_az': [8.582, 0.036]}, {'vdss_lombardo': [0.707, 0.006]}, {'half_life_obach': [0.511, 0.000]}, {'clearance_hepatocyte_az': [0.457, 0.013]}, {'clearance_microsome_az': [0.620, 0.007]}, {'ld50_zhu': [0.588, 0.000]}}