The tested ubiquitin correlations have no independent accuracy check.
Across three selected ubiquitin coordinate ensembles (1D3Z, 2K39, and 6V5D), this analysis compares equal-weight alignment with a diagonal variance rule at four per-coordinate variance floors. Adding residues 71–76 to the nominal 0.03 Å variance fit changed core dispersion by +0.0300651%, +0.0259508%, and −0.0518131%, with RMS correlation changes of 0.00601129–0.00696100. Equal-core versus variance-full fits differed more, and changing the floor from 0.01 to 0.3 Å had its largest tested effect in 1D3Z (RMS correlation change 0.125927; maximum 0.441019). All 60 fits converged and checks passed, but agreement across fits measures stability rather than accuracy: the coordinate files contain no reference correlation matrix. The 302 models are selected, related structural representations, not independent biological samples. Results apply only to the saved files and rules; no biological accuracy, confidence interval, p-value, or physical population was established.
ubiquitin · protein structure · structural alignment · residue correlation · variance weighting · coordinate ensembles · model sensitivity
References
3- usesUbiquitin tail inclusion effects depend on alignment weighting across three selected ensembles sci:sha256:4739203bc92ac6188014210e3c674f10fee53a165b0dfb4045ab0bf28d6d682a
I use the coordinate hashes, residue masks, diagonal variance rule, and four variance floors. I also use 12 reported spread values to check the calculation.
- doi:10.1371/journal.pcbi.0040043
- doi:10.1007/s10858-019-00288-8
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesEach model removal changes the 1D3Z correlation matrix more than tail inclusion at a 0.03 Å floor.
I use its 70-by-70 correlation definition and three printed tail-mask RMS changes. They define the comparison and check the reproduced baseline.