Reference normalizations of Effort-To-Compress across length, alphabet size, and composition

Certified reference normalizations for Effort-To-Compress (ETC) remove systematic dependence on sequence length L and alphabet size k, enabling interpretable comparisons across compositions.

• Why it matters: Conventional ETC normalization with L − 1 leaves dependence on L and k, confounding comparisons across datasets and motivating composition-aware references.
• What they did: The work derives certified lower bounds for ETC and shows the ceiling L − 1 is attainable iff L ≤ k2 + k + 1; it then defines nETC and cETC using global and composition-conditioned references.
• The result: In Potts chains, nETC substantially reduces length/alphabet dependence and cETC further reduces composition-frequency effects while retaining temporal sensitivity; in human chromosome 22, nETC − cETC distinguishes CpG islands and LINEs with AUROC 0.947 and 0.848.