The paper discusses in detail potentially new drugs discovery for prostate cancer disease by exploiting literature-derived knowledge and semantics. Using knowledge discovery methodologies is, recently, getting popular for all types of drastically growing data sources such as medical data. Although the authors argue that their methodology is groundbreaking and it sounds reasonably effective, the methodology could have been more practical and automated by reducing a manual intervention step and improving the path generation part in the whole process.
Summary:
The authors claims that bringing a new drug to the market is extremely expensive and time consuming because of scientific experiments and official approvals. The claimed approach exploits using semantic predications extracted …show more content…
Semantic predication extraction from SemMedDB
They only extract three types of predications: Gene-Cancer, Gene-Gene, and Drug-Gene. For each predication type, they use predefined set of predicates to generate reasonably meaningful pathways. Although these extracted predications might provide decent information to find potentially important pathways between target disease and potential drugs, they are not throughout for finding all possible connections.
3. Prostate cancer discovery pathways
The authors generate two different pathway schemas from the extracted predications. These schemas are "Drug - Gene - Cancer" and "Drug - Gene- Gene - Cancer". It is likely possible to hit some meaningful pathways with these schemas. However, this set-up does not guarantee that they are able find possible drugs for the target disease.
4. Physician selection of semantic predications
In this step, the authors try to filter the populated results by a physician review. Certainly they get some irrelevant paths and they need to prune them to obtain potentially essential paths. Since this step involves a manual operation, it is evident that the whole process is not fully automated.