Dear Rescomp,
I have DNA data (1-D spatial data) that I am trying to analyze for  
spatial autocorrelation. At specified distances along a piece of DNA  
I have polymorphism values (range: 0-1) and I am interested in  
finding out if values close to each other are more similar to each  
than than point far apart. A previous researcher used time series  
tools (autocorrelation and autoregressive models) to analyze such DNA  
data.

Our four different one dimensional spatial series  each have very few  
points (9-18 points). Not surprisingly, none of these four separate  
data sets shows significant autocorrelation. This is most likely due,  
at least in part, because of a true lack of autocorrelation. I also  
assume we have NO power to detect a pattern if it was there.

I tried using a Mantel test to determine if there were a spatial  
pattern. (The Mantel test determines the similarity of two matrices.  
In this case the matrices were all the pairwise euclidean distances  
between (i) true DNA distances and (ii) polymorphism values.  
Unfortunately, this test does not even reveal any pattern on  
different test data that reveal a significant autocorrelation.)

***I need the to find the most powerful way to test our hypothesis of  
proximity-similarity and references that can provide support the  
choice of such a method. Any ideas?***

I attached an example of our data in a CSV file with four columns and  
19 rows.
Col 1 is record number. I am interested in whether values of PI (col  
3) tend to be more similar if they originate from positions (kb, col  
2) that are close together.




Thank you!
Hank

Dr. M. Hank H. Stevens, Assistant Professor
338 Pearson Hall
Botany Department
Miami University
Oxford, OH 45056

Office: (513) 529-4206
Lab: (513) 529-4262
FAX: (513) 529-4243
http://www.cas.muohio.edu/~stevenmh/
http://www.muohio.edu/ecology/
http://www.muohio.edu/botany/
"E Pluribus Unum"