Biophysical Descriptors of Single Protein Domains

<jats:p> Quantitative relationships among geometric and surface-chemical descriptors of single-domain proteins remain insufficiently characterized, despite their relevance for interpreting structural datasets and modeling biomolecular interactions. Using a curated, non-redundant set of 614 single-domain proteins, we analyze the hydrophobic fraction of the solvent-accessible surface area ( <jats:italic toggle="yes"> SASA <jats:sub>H</jats:sub> </jats:italic> ), charged surface fractions ( <jats:italic toggle="yes"> SASA <jats:sub>+</jats:sub> </jats:italic> and <jats:italic toggle="yes"> SASA <jats:sub>−</jats:sub> </jats:italic> ), size measures, secondary-structure content, and shape descriptors derived from high-resolution meshes. <jats:italic toggle="yes"> SASA <jats:sub>H</jats:sub> </jats:italic> exhibits a narrow, reproducible range (0.45–0.65), whereas <jats:italic toggle="yes"> SASA <jats:sub>+</jats:sub> </jats:italic> and <jats:italic toggle="yes"> SASA <jats:sub>−</jats:sub> </jats:italic> follow similar Gaussian-like distributions but centered near 0.1, indicating conserved patterns of hydrophobic exposure and surface charge. Correlation analysis reveals strong interdependence among size descriptors and an inverse relationship between sphericity and surface roughness. A minimal, near-orthogonal set—mass, sphericity, <jats:italic toggle="yes"> SASA <jats:sub>H</jats:sub> </jats:italic> , net charge, and helical content—captures the dataset’s essential variation with minimal redundancy. This framework provides geometry-aware descriptors for analyzing large collections of predicted structures, including those generated by AlphaFold, and offers physically interpretable parameters for protein comparison, coarse-grained modeling, and the design of domains with tailored surface properties. </jats:p>