Questions&Comments



1If I want to predict ATC codes by SPACE, what information I have to provide?Asked by Liuzy(114.246.52.71) at 2014-06-16 07:39:11  Reply  Delete

To predict ATC codes by SPACE, now you have to provide the Pubchem_CID of the interested compound and its structure information represented by InChI (Please see the “Example” of the SPACE homepage). If you cannot obtain the Pubchem_CID of the compound, you can use other ID (it must be non-digital, because digital will be automatically thought to be Pubchem_CID) to represent your compound. But under this condition, SPACE predicts ATC codes for the query compound only based on two structural similarity-based features. The InChI structural information is compulsive, otherwise ATC codes cannot be correctly predicted! In future, more other kinds of IDs and structural descriptors will be supported.(Replied by liuzy)

2You stated that the enrichment analysis is based on query compounds’ predicted ATC codes ranked in the top xxx and meanwhile with probability_score larger than xxx. My question is that for a “classified” drug, are its known ATC codes considered in the enrichment analysis?Asked by Liuzy(103.242.203.207) at 2014-06-16 07:47:40  Reply  Delete

For the result of ATC code prediction, for a “classified” drug, we not only give its known ATC codes, but also give predicted ATC codes with probability_score greater than xxx and meanwhile ranked in the top xxx (not including its known codes!).
For the enrichment analysis, for a “classified” drug, we consider ATC codes ranked in the top N (including its known codes. that is, e.g. if N=5 and the number of its known codes is 3, we consider the 3 known codes and 2 predicted codes ranked in the top 2) and with probability_score greater than xxx.(Replied by liuzy)

3How we cite SPACE?Asked by Yu C(111.199.188.25) at 2015-09-05 03:02:09  Reply  Delete

Please cite: Liu Z, Guo F, Gu J, Wang Y, Li Y, Wang D, Lu L, Li D, He F. Similarity-based prediction for Anatomical Therapeutic Chemical classification of drugs by integrating multiple data sources. Bioinformatics. 2015, 31(11):1788-95.
Thanks a lot!(Replied by liuzy)


Leave your question or comment here