[R] find multiple mode, sorry for not providing enough information

[Jim Lemon]

Hi Ding,
While I was completely off the track in my first reply, the subsequent
posts make your problem somewhat clearer. The way you state the
problem suggests that the order of the values of "freq" is important.
That is, it is not just a matter of finding local maxima, but the
direction in which you approach those maxima is important. For
example. I might want to only identify maxima with at least four
monotonically increasing values preceding them and a decrease of at
least half the value of the maximum in the succeeding value. By
breaking down the problem into a set of criteria, these can be
implemented in a function that will search the values in one
direction, returning the locations of maxima that fulfil those
criteria.

Jim

On Mon, Mar 16, 2020 at 3:11 PM Yuan Chun Ding <ycding using coh.org> wrote:
>
> sorry, I just came back.
>
> Yes,  Abby's understanding is right.
>
> > tem4$Var1
>  [1]  1    3   4   5   6    7   8   9  10  11  12  13  14  15  16  17  18  20   21   22    23     24   25   31
> > tem4$Freq
>  [1]   1   2   5   5  10   4   4   8   1    1    8    8    2    4    3    1    2    1     1   138  149    14    1     1
>
> I have 2000 markers, this is just one example marker, the var1 is a VNTR marker with alleles 1, 3, 4 etc, a multi-allele marker; the corresponding frequency for each allele is 1,2 5 etc.  I want to convert this multi-allele marker to bi-allele markers by choosing a cutoff value; I would want the cut point to be allele 6 with frequency of 10, so allele 1 to allele 9 are considered as "short" allele, allele 10 to 31 as "long" allele;  then sliding to next rsing frequency peak, allele 8 with frequency of 8, etc.
>
> maybe those rising peaks are not really multiple modes, but I want to do this type of data conversion.  I want to first determine the number of modes, then convert input dat file into m different input files, then perform Cox regression analysis for each converted file. I am stuck in the step of find out m rise peaks.
>
> Thank you,
>
> Ding
>