Phenotypic insufficiency of local protein representations: fibre-preserving rearrangements of human C2H2 zinc-finger arrays reorder DNA-recognition residues under an invariant 25-spectrum
Abstract
A protein representation built from local sequence statistics assigns the same vector to all sequences with the same multiset of k-mers, the spectral fibre. Whether this blind spot matters depends on the phenotype. For C2H2 zinc-finger proteins, whose fingers read DNA in order, it can. For k at least the length of a finger, the multiset of fingers is the same for every member of a fibre; only their order can change, and only across boundaries formed by repeated (k−1)-mers. The conserved TGEKP linker is far too short to form such a boundary at k = 25. We analysed all 17 reviewed human C2H2 zinc-finger proteins with a non-trivial 25-fibre. Fibre-mates were drawn as random Eulerian trails of the de Bruijn graph of order 25, fingers were located by the C-X(2-4)-C-X(12)-H-X(3-5)-H motif, and each finger was summarized by its recognition residues at helix positions −1, 2, 3 and 6. Of 1,178 distinct fibre-mates, 883 (75%) place different recognition residues at some array position. In all 1,178 the finger multiset is unchanged and a window-25 convolutional encoder gives identical output. In ZNF91, a KRAB zinc-finger protein that represses SVA and L1 retrotransposons, one fibre-mate changes the recognition residues at 17 of 31 array positions; in ZNF721, one changes 17 of 23. KLF18 is a negative control: its fibre is large, but none of its 285 mates reorders its three fingers. Under any finger-additive recognition model, the predicted binding site changes exactly when the ordered recognition residues change. Binding was not measured; statements about binding sites refer to such models. This record contains the paper (PDF and LaTeX source), the code, the UniProt snapshot, all results and the figure. Companion papers: doi:10.5281/zenodo.22950095, doi:10.5281/zenodo.22959211, doi:10.5281/zenodo.22960439, doi:10.5281/zenodo.22963882, doi:10.5281/zenodo.22966020. Code: https://github.com/Ruqing1963/zinc-finger-order
// Source
Authors: Zhengyi Chen, Ruqing Chen
Institutions: Guilin Medical University