calcPCClocks_Accel fix

[DanielEWeeks]

The calcPCClocks_Accel function did not work for me on your example data until after I changed this line

#   DNAmAge[,paste0(i,"Resid")] = resid(lm(DNAmAge[,i] ~ DNAmAge$Age))

to

    DNAmAge[,paste0(i,"Resid")] = resid(lm(eval(parse(text=paste0("DNAmAge$",i))) ~ DNAmAge$Age))

The original code generated this error message:

Error in model.frame.default(formula = DNAmAge[, i] ~ DNAmAge$Age, drop.unused.levels = TRUE) : 
  invalid type (list) for variable 'DNAmAge[, i]'
Called from: model.frame.default(formula = DNAmAge[, i] ~ DNAmAge$Age, drop.unused.levels = TRUE)

I believe this is because DNAmAge[, i] is a tibble instead of the expected numerical vector.

head(DNAmAge[,i])
# A tibble: 6 x 1
  PCHorvath1
       <dbl>
1       57.4
2       52.7
3       49.9
4       50.5
5       48.5
6       54.4

Thanks,
Dan Weeks

[GuoFengWang]

I used the clcPCClocks_Accel to calculate the data in the example and found that the PCGrimAge is greater than the actual age, but the calculated epigenetic age acceleration is a negative value. Why is this? Shouldn't it be that when the epigenetic age is greater than the actual age, the epigenetic age acceleration is positive, and when the epigenetic age is smaller than the actual age, the epigenetic age acceleration is negative? @DanielEWeeks @kthrush @albertchen42

[GuoFengWang]

Here is my clcPCClocks_Accel result from example data:

[audreylk]

@GuoFengWang - I think it's probably due to the fact that the age acceleration is not just a difference between the epigenetic age and the chronological age, but is a residual of the epigenetic age on chronological age for the whole population sample. So a very high shift in the intercept of this regression for the whole population could mean that you get a negative residual for an epigenetic age that is greater than the chronological age, if that difference is less than the overall population shift. It just generally looks like (from these 10 samples) the PCGrimAge is pretty high for everyone overall, so I think that's what happened here.