There are various studies in which they measure the amount of melatonin in subjects' saliva, after exposing them to different wavelengths of light at the same ... well, in this case it's energy rather than lux:
https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.96...> Two long wavelengths, 595 nm (amber) and 660 nm (red) and three shorter wavelengths, 470 nm (blue), 497 nm (blue/green), and 525 nm (green) were compared with a no-light control condition. Light was administered via a portable light source comprising two light-emitting diodes per eye, with the irradiance of each diode set at 65 µW/cm 2. [...] Half-hourly saliva samples were collected on the evening prior to the first light pulse and the evening following the second light pulse. The time of melatonin onset was calculated for each night and the difference was calculated as a measure of phase advance. The shorter wavelengths of 470, 495 and 525 nm showed the greatest melatonin onset advances ranging from approximately 40–65 min while the longer wavelengths produced no significant phase advance.
Looking at a second source for lux at different wavelengths... https://www.circadiansleepdisorders.org/info/lux.php
It says light sensitivity peaks at 555 nm, and the graph looks mostly symmetrical (though the red side is slightly higher than the mirrored blue side). If it were all about perceived brightness, then I think we'd expect amber light to have a stronger effect than blue-green or blue, but that was not observed.