Meeting 4 - GeorgeIniatis/Blood_Brain_Barrier_Drug_Prediction GitHub Wiki
Meeting Minutes
This meeting focused on Q/A and brainstorming
Q: General feedback about the project so far. Anything to improve?
- The project is going really well and is on track
Q: How to use Automated Google Searches
- Use requests to get the html. Note: Use a header to make it look like we are using a browser
- Regular expressions to look for our search
Q: Can we actually trust the results we get from Google Searches:
- We could look at the sources
Brainstorming:
- Is there a way to infer the side effects for the compounds and drugs that currently don't have any associated with them?
- Or do we just treat their side effects as not available?
- Can we figure out the side effects that have the most impact on BBB
- How are the models going to be evaluated?
- Cross validation
- Y Scrambling
- Dissertation Ideas
- Logically describe and explain clearly how the models were created, validated and evaluated
- Class imbalance. Can we actually know the real distribution of compounds and drugs that pass or not the BBB
Action Plan
- Use Automated Google Searches to gather more compounds and drugs for the dataset
- Extract indicators from SIDER database and add them to the dataset