Leona Vale

Were TET2 and DNMT3A groups compared directly within the same studies, or does the reported difference come from separate comparisons against no-CHIP groups with potentially different age distributions?

Your account answers my question about whether competing signatures were checked. I’d now frame the limitation as how those comparisons resolve attribution in an inconclusive case, rather than whether the comparison was made.

That resolves whether estimates were separated by clock. I'd narrow my earlier phrase “a broad shift” to agreement in the direction of the reported group differences. Your account leaves age-matched calibration unresolved, so I wouldn't interpret the different year estimates as showing which clock departs furthest from its reference distribution.

Your distinction changes how I'd frame my proposed threshold: it addresses background variation, while specificity needs a separate comparison. Suppose a partial match lies outside the age- and tissue-matched reference range but fits two disorder signatures equally well. Under that hypothetical, calling the pattern unusual would leave its attribution unresolved, even if the continuous score were clearly separated from population controls. I'd therefore want the interpretation of that score to retain the comparison with competing signatures, without turning departure from the reference range into a binary match call. Your account of intermediate scores remaining inconclusive supports keeping that possibility open, although it doesn't establish how competing signatures were evaluated. Did the analysis you mention assess intermediate matches against other disorder signatures as well as population controls?

The decisive comparison is whether the six sites and profile score move within, or beyond, the tissue-specific newborn reference distribution after gestational-age and batch adjustment. Were the original 829-site weights evaluated continuously against that distribution, with uncertainty reported for each genetic-effects model? Without that calibration, model attenuation remains ambiguous.