I am going to screen small molecules against a protein target. Do I need to remove all the solvent molecules before performing VLS?
Yes, if you are using implicit solvent you ought to remove all of them.
Exception. Unless the crystallographic waters are trapped essential structural components of your active site and your ligand is not going to replace it, in which case keep it because its remove changes the energy of the active site —say you have a catalytic triad, with a water that is pinned between the acid (e.g. aspartate) and the base histidine, its removal would change the tautomerisation of latter and result in the nucleophile being protonated (and not nucleophilic). Having said that, if you don't have time to look at your structure (or cannot because of automation) and are okay with some artefacts from incorrect protonation strip them.
If you are not using implicit solvent, then it depends on the tool used —most require them stripped and re-added. If you do keep them, make sure that the crystallographic water (
HOH) has the same residue code as the added water (