Your minimap is just another buffer you render to. Same as the screenbuffer, it holds no information other than the pixel rgba values.
What you need is a translation between minimap coordinates to worldspace coordinates. So that e.g. the, say, 400×400 px the minimap takes up on your monitor, map to the 1000×1000 area of worldspace you display on that minimap.
So with the above values we select minimap-pixel 200/100, we get world-space 500/250.