Saturday, March 15, 2014
Calculating the lag phases in growth curves by minimizing sum of squared errors
I studied '1471-2180-11-140-s1.m', a matlab script that can estimate lag phases from growth curves. This program can 'align' growth curves using the matlab optimization procedures. I used Octave 3.8.0 with GUI and found it useful.
My plan to is to use R optim() to minimize the sum of errors. The differences of lag phases are the indices in the vectors. So, somehow, I need to re-index the vectors. Maybe I can use the a small threshold and put NA to very small values. I will then have two vectors, X1 and X2. The sum of errors can be defined as
SumOfErrors = sum( (X1[1:lengthOfX2] - X2[delta:END2] )^2 )
This is basically the global alignment problems in sequence alignment.
1471-2180-11-140-s1.m from the Xavier group.