Principal Component Analysis of Epigenetic Age (Morgan Levine)

    In one of today's ARDD 2022 Meeting presentations (8/31/2022), Morgan Levine mentioned that her group at Yale had developed a new method of eliminating the "noise" in using CpG methylation of DNA to determine epigenetic age from a blood or tissue sample.  She showed some graphs of improved epigenetic age estimates with and without the technique, but she wasn't very specific about what was being done to do noise elimination.  Doing an online search, I found an interview (LINK) in which she went into more detail and provided a reference (LINK).

    Basically, the trick is to do a Principal Component Analysis on the large quantity of CpG DNA methylation data from an array chip, and then do the correlation analysis between the derived principal components and subject's calendar age.  This has the effect of removing outliers and eliminating the noise variation of around 5 or so years in Horvath and similar epigenetic clocks and reducing it to about a year.  Here's a figure showing the improvement (PC) with various clocks.

    The question now is when the commercial firms that will analyze a blood or saliva sample and provide epigenetic age will start providing analyses that employ this new technique.