When trying to decide which additive might be useful during crystallization, try the following.
If one has crystals and wants to try using additives to improve the crystal size or quality go back to the initial crystallization screening plates. Review the plates, looking for conditions where no precipitate nor crystals were/are observed. Now, review the results, looking for a common reagent ingredient present where no precipitant or crystals are found. For example, one might find that all drops with isopropanol remained clear. One could then try adding isopropanol to the current crystallization condition to see if the isopropanol could improve the crystal size and/or quality. If one observes a difference in the crystal in the presence of isopropanol, then one might consider evaluating other additives in the class of alcohols such as ethanol, methanol, tert-butanol and others.
If one has no crystals, but plenty of precipitate, phase separation and clear drops, follow the above analysis and try adding the common reagent ingredient found in clear drop to those drops which contained precipitate or phase separation. It is possible this agent could improve or at least manipulate sample solubility.
The above tip was submitted from Jarmila Jancarik from the laboratory of Sung Ho Kim at the University of California Berkeley. Thank you Jaru!
When using this approach it might be reasonable to discern between concentration independent and dependent precipitation when trying to decide which agent to pursue as an additive. Try evaluating the concentration independent agents first and then look at the other agents if sample quantity permits. For example, if one observe precipitate in 15 to 40% PEG but not in 5 to 10% PEG, it might simply be a concentration. However, if one observes no precipitate in 45 to 60% v/v MPD one could guess that MPD is a reasonable agent to evaluate as an additive.