The protein cryptochrome is thought to be the magnetosensor in migratory birds, i.e. the protein responsible for the avain magnetic compass. This ability to sense magnetic fields is provided by a radical pair that can be generated inside cryptochrome after activation of the protein, and cryptochrome is thought to be activated by absorption of light. Experiments have shown that light of specific colors (specific wavelengths) is required for birds to utilize their magnetic compass sense, and this is a strong indication that certain molecular excitations inside cryptochrome must be triggered by light prior to generation of the radical pair. Therefore, the photochemistry of cryptochrome is of crucial importance for the magnetic compass sense.
You can read more about the avian magnetic compass here.
Photoabsorption by flavin adenine dinucleotide (FAD)
Many types of cryptochrome does not only consist of amino acids, but have smaller molecules embedded inside the peptide chain, and many cryptochromes are known to possess a specific molecule, namely an FAD molecule which has some interesting photochemical properties. FAD is well known to absorb light, mainly in the blue part of the visible spectrum, and it has been shown that the excited state of FAD can lead to formation of a radical pair inside cryptochrome. One might, therefore, conclude that FAD is solely responsible for the generation of the magnetosensitive radical pair, and that this is the full story. Unfortunately it is not that simple - studies such as [Muheim et al.; J. Exp. Biol., 205, 3845-3856 (2002)] and [Wiltschko et al.; J. R. Soc. Interface, 7, 163-177 (2010)] have shown that migratory birds may utilize their compass sense when blue light is available, but not when only red light is available, in full agreement with the absorption spectrum of FAD. But as shown in the same two studies, things get a bit more fuzzy when considering green light; birds were still able to navigate while exposed to only green light, even though FAD not really absorbs light in that range. A possible explanation of this conundrum is that FAD gets affected by its immediate molecular environment, i.e. that the protein in which FAD is embedded may affect FAD in such a way that FAD suddenly is able to absorb green light.
A computation of the absorption spectra of a variety of different cryptochromes were carried out in [Nielsen et al.; J. Phys. Chem. Lett., 9, 3618-3623 (2018)], taking the molecular environment into account. Since the total molecular environment is too large to be account for using quantum mechanical means, a so-called Polarizable Embedding (PE) potential was used to represent the environment, while the flavin part of FAD was treated quantum mechanically. The PE potential represents the environment using a second-order multipole expansion of the charges (i.e. monopoles, dipoles, quadropoles), together with a set of polarizabilities, such that the environment could be polarized by the so-called quantum region (the flavin part of FAD). Such a representation of the environment is not as good as a full quantum mechanical treatment, but is a state-of-the-art method invented for such purposes. The absorption spectra calculations were performed using damped response theory, which is a method for obtaining the spectrum of electronic excitations in a molecule - but note here that only electronic excitations are taken into account, and not coupling to the vibronic modes of the molecule. This effectively means that the "true" spectrum may be slightly different due to vibrations in the molecule. The calculated spectra were, however, averaged over several structures taken from a molecular dynamics simulation, so the random thermal motion of the system was accounted for in this way.

The calculated absorption spectra of FAD embedded inside cryptochrome shows no absorption in the green-light region of the electromagnetic spectrum, and the vibronic couplings missing from our calculations are very unlikely to make such a large difference as would be necessary for FAD to absorb green light. It was therefore concluded in [Nielsen et al.; J. Phys. Chem. Lett., 9, 3618-3623 (2018)] that something else must be going on inside cryptochrome - perhaps that FAD is not the only small embedded molecule, but that a secondary photoreception may exist within the type of cryptochrome that may be responsible for avain magnetoreception. Secondary photoreceptors are seen in some cryptochromes already, so it is not impossible that another small molecule may reside there, but have yet to be discovered.
The calculated spectra can be seen below.
