Total TCMs:490
Total TCM ingredients:12,629
Chemical Drugs:1,241
Predicted relationships:169950

Brief Introduction

Generally, the 1.0 version of TCM-PTD used structual-similarity comparison to predict potential targets for TCM ingredients.

The TCM database elements map are linked like this:

TCMs<->TCM ingredients<->(Chemical drugs<->)Targets

Drugs or ingredients were presented by QSAR descriptors (DRAGON software), and target proteins were presented by Pseudo Amino Acid Composition (PseAAC).

The golden positive datasets was constructed using approved drugs -target pairs collected from DrugBank, and randomly generated same number of drug-target pairs as negative datasets. a Random Forest predicitng model was then built upon these pairs.

The 8777 ingredient from TCM related with 4147 targets were then test by the predicting model. Pairs with high prediciting score were considered as potential targets for this ingredient. Pairs with predicting score >0.9 were picked, and at least 10 potential target (even the predicting score <0.9) were selected for candidates.

In current version (1.0) , TCM ingredients were compared to only FDA approved drugs (1401). and 8777 ingredients from 387 TCM collected by Chinese Pharmacopeia (2010 vol.1) were predicted, involving 1254 drug targets.

We used this comparison result to construct TCM-target relationships. Ingredients From TCM are freely chosen by users as different pre-treatment of TCM may cause to different ingredients remained.