We conclude that new IDR-specific features and paradigms are needed to accurately classify disease mutations within those regions. Disorder-to-order transitions predicted by AlphaFold3 and motif predictions (e.g. by SHARK-capture) may aid structure-based variant interpretation.

The distance to the second Methionine discriminates between pathogenic and benign variants associated with the N-Methionine site.

Mutations at the N-Methionine site are prevalent. Look at the surprisingly high peak at position 1.

Most mutations in disordered regions are benign but ~12% of pathogenic mutations fall within a disordered region.

The correlation between the disorder predictors is low.

It depends which disorder definition (score) is used. Only 7% of residues are consistently predicted as disordered across five different disorder prediction tools. flDPnn under predicts disorder compared to other methods.

State-of-the-art VEPs tools have lower sensitivity but higher specificity for variants in disordered regions.

Proteome-wide predictions of motifs in S. cerevisiae are available. Matches to ELM motifs, UniProt motifs, and mutagenesis sites (UniProt) are enriched in SHARK-motifs compared to the random control. An example output: SHARK-capture score plot for Sic1.

SHARK-capture identifies a highly conserved motif (in pink) amongst Ded1p orthologs which regulates ATPase activity. Interestingly, AlphaFold2 predicts close proximity between the motif and the helices core.

SHARK-capture is available as a Python package (pypi.org/project/bio-...) and on git (git.mpi-cbg.de/tothpetroczy...) where we provide bash scripts for parallelization on HPC and a colab notebook colab.research.google.com/drive/1l_hNY...