About Statistical Calculations for Scores

Question

[aaaburak20]

I have 255 genetic associated genes from Target Validation Platform and their scores for genetic associations. Also I have relationship scores between different genes from STITCH. I will decrease the number of genes in Target Validation Platform by using score distribution. Then I want to construct some networks about Alzheimer's Disease by using these genes and also I want to include these scores from TVP and STITCH. But what kind of function can I use to combine these different types of scores to build a network and find the most relavant gene about AD? If there is something wrong about my idea and procedure, could you please help me to correct it?

Thanks.

[Denise CS]

What is your rationale for combining them together? I'm not sure if I'd combine the scores from our Open Targets Platform and STITCH. The scoring system between those two is rather different. For STITCH for example the scoring is based on text mining only, whereas Scoring target-disease association by us relies on other datasets besides text mining, e.g. genetic association, RNA expression. The text mining approach does create lots of false positives, so I'd try to focus on the data based on experimental approaches only to try to build my network. If you'd rather go ahead with the combining approach, you may want to explore the literature e.g. A new scoring system in Cystic Fibrosis: statistical tools for database analysis – a preliminary report for some ideas or try taking this to statistical forums such as TalkStats or Statistical Forum. There are plenty of those out there.