Extended Data Fig. 2: Distribution of per-residue lDDT-Cα within eight pLDDT bins.
From: Highly accurate protein structure prediction for the human proteome
This represents an alternative visualization to Fig. 1a that does not sample the data. It uses the recent PDB dataset (Methods), which is restricted to structures with a reported resolution of <3.5 Å (n = 2,756,569 residues). Residues were assigned to bins of width 10 based on their pLDDT (minimum, 20; maximum, 100). Markers show the mean lDDT-Cα within each bin, while the lDDT-Cα distribution is visualized as a Matplotlib violin plot (kernel density estimate bandwidth, 0.2). The smallest sample size for the corresponding violin is 5,655 residues for the left-most bin.
