K-mer Conservation Analytics
DOE Phase II Objective 1 — cross-organism k-mer conservation scoring
What do these k-mer views mean? — plain-English guide
A k-mer is a short fixed-length “word” of a protein sequence — here, every run of k = 8–12 consecutive amino acids. A 300-residue protein contains ~300 overlapping 8-mers.
Conservation = how widely a k-mer recurs across different organisms. A k-mer seen in many species is evolutionarily conserved — usually because it does something important (a binding site, catalytic motif, structural core), so evolution hasn’t let it drift. Rare / organism-specific k-mers are the opposite.
Why EASME cares: conserved regions are evolutionary constraints. When designing or mutating proteins you generally want to preserve highly conserved motifs and target the variable regions — so conservation scoring tells you where change is safe vs risky.
Glossary
| Score / cons. score | How conserved one k-mer is across the catalog (higher = more widespread). Blends abundance and phylogenetic spread. |
| F_kmer | A whole-protein score = the mean conservation of all its k-mers. High = the protein is built largely from widely-shared motifs. |
| w_abund | Abundance weight — how often the k-mer occurs overall. |
| w_phylo | Phylogenetic weight — how spread across the tree of life the occurrences are (rewards distant organisms over many copies in one). |
| Organisms | How many distinct organisms contain the k-mer. |
| Universal | K-mers present in (nearly) every organism — the conserved “core vocabulary.” |
| Match % | For one protein: the fraction of its k-mers found anywhere in the conservation index. |
| Specificity | (Clade view) within-rate − cross-rate. 1.0 = k-mer is exclusive to the chosen clade; 0 = equally common inside and outside. |
| Exclusive | K-mers found in only that one organism (not shared with others). |
Scores look tiny (e.g. 3.5e−7) because they’re normalized across a ~766K-k-mer index — read them relative to each other, not as absolute values.