This is an age of incredible access to protein structures, both experimental and predicted, and as scientists, our decision making and hypothesis building is shifting to incorporate more and more structural information. Why? Because structures can help us develop informed hypotheses around mechanism of action, guide small molecule and protein binder design, and engineer rationally designed variants to push our protein towards a desired outcome.

We all understand the concept of over-interpretation of data - trying to draw conclusions that are far too detailed or nuanced given the data quality or experimental parameters. And the same applies to protein structures! There are several structure refinement statistics that help us understand structure quality, and refinement resolution is a critical first parameter to consider before you download that PDB file and design away.

Refinement resolution tells you how precisely atomic positions were determined in a structure. Lower numbers mean higher precision, and often that means greater detail can be reliably extracted.

𝗛𝗼𝘄 𝗱𝗼𝗲𝘀 𝘁𝗵𝗶𝘀 𝗽𝗹𝗮𝘆 𝗼𝘂𝘁?

For small molecule design, structures at < 2.0 Å allow you to clearly see ligand geometry, hydrogen bond networks, and even water molecules. A level of detail that greatly helps drug design. For structures with resolution between 2.5-3.0 Å, ligand poses can require some interpretation and there can be more ambiguity in binding mode. And > 3Å? Stop there. Get a new (higher resolution) structure.

For protein engineering, structures at < 2.0 Å provide the resolution necessary to assign sidechain placement and discern residue-residue interactions within and between proteins. At > 2.5 Å, flexible or solvent-exposed residues can be more challenging to observe and similar to small molecule design, the lower the resolution goes, the more you're guessing rather than observing.

𝗪𝗵𝗮𝘁 𝗰𝗮𝗻 𝗰𝗮𝘂𝘀𝗲 𝗹𝗼𝘄 𝗿𝗲𝘀𝗼𝗹𝘂𝘁𝗶𝗼𝗻 𝗼𝗳 𝗮 𝗽𝗿𝗼𝘁𝗲𝗶𝗻 𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲?

Low resolution can stem from intrinsic flexibility or disorder within a protein, so sometimes low resolution is also teaching you about a protein’s structure, or lack thereof. In the context of X-ray crystal structures, low resolution can also be a result from suboptimal protein packing, weak or noisy crystal diffraction, or radiation damage during data collection. All aspects that we can experimentally optimize to achieve the level of resolution necessary to match the level of interpretation that we’d like to achieve!