Please show the spectra so that we can see what the difference is (or just describe the difference in detail, if you would prefer not to show the spectra) and can attempt to interpret it.

Other than that, the most fundamental suspect would be the presence of water, which is a vast and broad signal that can overlap virtually anything in the 1700-4000 cm-1 range.

I agree with Tomasz Jarosz . Something that may also factor in, is that powder spectra comply mostly with the Bouguer-Beer Lambert law Article Employing Theories Far beyond Their Limits-The Case of the (...

Article The Electric Field Standing Wave Effect in Infrared Transmis...

From the above description it looks to me (I'm guessing as it is not stated clearly) that you're measuring the FTIR transmittance spectra by sending the IR beam through your samples, i.e. in the transmittance measurement mode in your spectrometer. If so:

You could also try to measure the FTIR spectra by utilizing an attenuated total reflectance (ATR) accessory, e.g. with a single-bounce diamond crystal as the ATR crystal. Using an ATR setup is a fast and simple way to measure most samples you encounter, both solid state and liquid samples.

Do all the solids dissolve in water? If they do not, then you will see mostly the solvent as Mr. Jarosz states. If the solids all dissolve, then it will be a matter of solvent interference, concentration of the dissolved solids, and possible hydrogen bonding effects on the protein which could cause it to unfold. I like the idea of attenuated total reflectance (ATR) as suggested by Mr. Jelle. This will reduce or eliminate overly-strong absorbances.