Polypharmacology Browser 2 (PPB2)
Home
Tutorial
FAQ
Contact
Draw or paste your query molecule here:
(Click here to load test compound)
Predict targets from compound - protein targets associations in ChEMBL22 using one of the following methods
Nearest neighbor search with:
Extended Connectivity fingerprint ECfp4
NN(ECfp4)
Shape and Pharmacophore fingerprint Xfp
NN(Xfp)
Molecular Quantum Numbers MQN
NN(MQN)
ECfp4 Naive Bayes Machine Learning model produced on the fly with 2000 nearest neighbors from:
Extended Connectivity fingerprint ECfp4
NN(ECfp4) + NB(ECfp4)
Shape and Pharmacophore fingerprint Xfp
NN(Xfp) + NB(ECfp4)
Molecular Quantum Numbers MQN
NN(MQN) + NB(ECfp4)
Naive Bayes machine learning model with entire dataset using:
Extended Connectivity fingerprint ECfp4
NB(ECfp4)
Deep Neural Network model with entire dataset using:
Extended Connectivity fingerprint
DNN(ECfp4)
Best performing methods are shown in bold. Please refer to manuscript for more details. Shortname is shown for each method.
Recall and Precision statistics considering top 10 predictions
Method
Recall (%)
Precision (%)
NN(ECfp4)
86
13
NN(Xfp)
81
12
NN(MQN)
76
11
NN(ECfp4) + NB(ECfp4)
76
41
NN(Xfp) + NB(ECfp4)
81
21
NN(MQN) + NB(ECfp4)
78
24
NB(ECfp4)
80
14
DNN(ECfp4)
82
21
How to cite
The Polypharmacology Browser PPB2: Target Prediction Combining Nearest Neighbors with Machine Learning. M. Awale and J.-L. Reymond, 2018, ChemRxiv, doi.org/10.26434/chemrxiv.6895646.v1
×
Error! missing input data
© copyright 2021 | all rights reserved.