Perhaps do cross validation among all expression stage groups and output another similarity metric (hence the score is weighted equally among expression stage groups instead of on each individual expression stage).
Create a specificity score that compares the correlation value from the selected expression stages compared to all expression stages. A high specificity would be a high correlation among the selected expression stages, and a low correlation among all the expression stages.
-For example, correlating CG1597 among embryogenesis has CG2867 as the most correlated gene. Although CG2867 is not even top 100 when correlating CG1597 overall. Therefore CG2867 would have a high specificity for embryogenesis.
Perhaps do cross validation among all expression stage groups and output another similarity metric (hence the score is weighted equally among expression stage groups instead of on each individual expression stage).
Create a specificity score that compares the correlation value from the selected expression stages compared to all expression stages. A high specificity would be a high correlation among the selected expression stages, and a low correlation among all the expression stages. -For example, correlating CG1597 among embryogenesis has CG2867 as the most correlated gene. Although CG2867 is not even top 100 when correlating CG1597 overall. Therefore CG2867 would have a high specificity for embryogenesis.