Skip to main content

Archived Comments for: Aging of blood can be tracked by DNA methylation changes at just three CpG sites

Back to article

  1. Comparison with the epigenetic clock by Horvath 2013

    Steve Horvath, Univ of California, Los Angeles

    18 February 2014

    Weidener et al (2014) present an age predictor based on human blood methylation levels that only uses 3 CpGs. Last year, I published an age predictor (referred to as epigenetic clock) that works well in most human cells/tissues/organs (Horvath 2013, PMID: 24138928) but it makes use of 353 CpGs. Given that a sparse predictor has obvious practical advantages, readers may be interested in learning how the epigenetic clock compares to the predictor by Weidner et al.

    To provide a fair comparison, I applied both predictors to the test sets mentioned in Figure 2 of (Horvath 2013). The comparison is fair because the epigenetic clock was not constructed (trained) on these test data sets.

    A Figure that shows the results of the comparison can be found here

    http://labs.genetics.ucla.edu/horvath/htdocs/dnamage/weidener2014/FigureComparison.pdf

    or on my webpage

    http://labs.genetics.ucla.edu/horvath/htdocs/dnamage/weidener2014

    The Figure shows that the predictor by Weidener leads to a good correlation with chronological age (r=0.72 across all test sets, Figure panel A) but to an unacceptably high median absolute error (13 years). When restricting the analysis to whole blood, the correlation is slightly higher (r=0.76, panel D) but the median error of 11 years remains high.

    By comparison, the epigenetic clock leads to a very high correlation (r=0.96, panel M) with age across the test set data and a very low median error=3.6 years. In all considered tissues and cell types, the epigenetic clock greatly outperforms the predictor by Weidener in terms of age correlation and median error (Figure).

    My analysis has two limitations. First, Weidener et al mention that their predictor works best on pyrosequencing data. I could only evaluate these two predictors on Illumina array data. A future comparison could try to compare both predictors on pyrosequencing data. Second, Weidener et al use a CpG site upstream of cg17861230. Since I only had Illumina array data, I had to use cg17861230 in my implementation of their predictor. Specifically, I implemented the predictor by Weidener as follows

    Predicted age=38.0-26.4*cg02228185-23.7*cg25809905+164.7*cg17861230

    Despite these limitations, I think these results strongly suggest that sparsity comes at a cost in terms of accuracy and in terms of applicability to other tissues.

    Figure: Evaluation of the predictor by Weidener et al 2013 and the epigenetic clock in test data sets

    http://labs.genetics.ucla.edu/horvath/htdocs/dnamage/weidener2014/FigureComparison.pdf

    Each scatter plot shows how chronological age (y-axis) relates to predicted age (x-axis). The caption of each plot reports the median absolute error and the correlation between predicted and true value. Points are labelled and colored by data set as described in Horvath 2013.

    A-L) The first two rows show how the age predictor by Weidener performs in different tissues/fluids/cell types. Note that the predictor leads to a moderate correlation but an unacceptably high median error even in blood tissue.

    M-V) The last two rows show how the epigenetic clock (Horvath 2013) performs on the same data sets. Note that it consistenly outperforms the predictor by Weidener in terms of age correlation and median error.

     

    Competing interests

    I am the author of the article Horvath (2013) and developed the epigenetic clock. Anybody can check my claims by using the software on my webpage: https://dnamage.genetics.ucla.edu/

     

  2. Response to comment "comparison with the epigenetic clock by Horvath 2013"

    Wolfgang Wagner, RWTH Aachen University Medical School

    27 February 2014

    Steve Horwarth has compared the method described in this article (Weidner et al., PMID: 24490752) with his recently published age-predictor (Horvath 2013, PMID: 24138928). This comparison is very interesting and helpful. We are grateful for this effort.

    As pointed out by Steve the comparison has two technical limitations that may contribute to higher median absolute error with our model:1) The multivariate model of our Epigenetic-Aging-Signature was trained on pyrosequencing data. There may be some technical deviation when using this model on Illumina array data although absolute beta-values should be similar; 2) The multivariate model of our Epigenetic-Aging-Signature uses a CpG site not presented on the Illumina microarray.

    Despite these facts, analysis of 353 CpGs (further CpGs may be used for normalization) can certainly provide better correlation with chronological age than only 3 CpGs. Furthermore, a predictor with many CpG sites can improve the accuracy when dealing with heterogeneous tissues. However, the rational of our predictor was not only to provide best correlation with chronological age but to provide a very simple, cost effective and robust tool which is applicable for a broad range of scientists. There is no need for microarray analysis or any bioinformatics knowledge. The higher deviations from chronological age may even better account for aspects of biological aging - but this deserves further validation in the future.

    In fact, we have also tested the Age-Predictor of Steve: it works indeed very well and we would like to
    encourage readers to give it a try - if Illumina array data are available (Horvath 2013, PMID: 24138928).

    With best regard,

    Carola Weidner, Qiong Lin, Wolfgang Wagner

    Competing interests

    We are coauthors of the manuscript "Aging of blood can be tracked by DNA methylation changes at just three CpG sites" Genome Biology 2014, 15:R24

Advertisement