DNA-R1B1C7-L Archives

Archiver > DNA-R1B1C7 > 2012-01 > 1326005059


From:
Subject: Re: [R-M222] FW: Calculation of a correlation coeficient
Date: Sun, 8 Jan 2012 01:44:19 -0500 (EST)


In a message dated 1/7/2012 4:39:15 A.M. Central Standard Time,
writes:
In order to illustrate, I need someone who is able to calculate, in Excel,
what Bill believes are correlation coefficients.

John, I vaguely remember that you played around with a few such correlation
coefficients, changed a few marker scores here and there to check the
effect
of a single mutations at various markers.

I checked my old notes. It's really pretty simple. There are two ways
to do it, the first is to use the CORREL function in Excel.

The syntax is:

CORREL(array1, array2)

You can compare any two marker strings of any length using this.

=CORREL(C1:AM1, C2:AM2)

Just enter the above formula with the appropriate column and row numbers
for each marker string.
Then you can change the markers and watch the correlation coefficient
change.

Bill does a conversion to RCC (revised correlation coefficient)

1/correlation efficient -1 X 10,000

=(1/.0988545-1)*10000

That results in something more usable like 15.5

You can also use the data analysis toolkit in Excel to compare more than
two strings at the same time. This toolkit does not come installed with
Excel - you need to install it from the installation disks. If you run that
you'll get a half matrix of correlation coefficient numbers for every
marker string in the spreadsheet range you specify. Bill then copies and pastes
the half matrix to form a full matrix which is identical to the matrix
used by the McGee utility.

Bill would then re-arrange the resulting full matrix to group similar
numbers together and said he could read the junction lines in the matrix. I
don't know how he did this. Never got that far with it.



John.



This thread: