Is there a way to rank the results of a linear model applied across thousands of values? For e.g. I have a table of 6004 rows, each corresponding to a gene and the corresponding output of linear regression model (I used the lm function in R):
gene estimate model.rSquared p.value
1 Slc25a4 17.442867 0.1954532 0.001654828
2 Mettl21c 12.104128 0.1780214 0.002817559
3 Pvalb 11.449275 0.1242439 0.014005302
4 Mybph 9.789242 0.1061216 0.023853234
5 Actn3 9.736698 0.1454820 0.007476086
6 Mybpc2 9.637181 0.1031207 0.026049394
I wanted to know if I can somehow rank these values using the combination of p-value, r-squared and estimate (absolute value of estimate to be precise)? Currently I am using a very naive approach of sorting the results first by p-value (low -> high), then by correlation (high -> low) and then by abs(estimate) (high -> low).
Thanks
