Closed wkingc closed 10 years ago
Wade,
Good find!
You have correctly identified a possibly-old and subtle bug. I imagine the dist
call was originally intended to be a data class coercion, not a call to calculate euclidean distance. Wow that's a bad one!
I have updated the github devel version, link should post here any second...
Thanks for the great catch! Always very appreciated, especially just-before a big BioC snapshot update :+1:
Hi Dr. McMurdie,
I am glad I could help. I am also really happy to have phyloseq, so anything to help make it better is good for me too ☺.
Thanks,
Wade
From: Paul J. McMurdie [mailto:notifications@github.com] Sent: Tuesday, October 07, 2014 9:27 PM To: joey711/phyloseq Cc: Copeland, Wade K Subject: Re: [phyloseq] Distance with method "dpcoa" inconsistant with results from ade4 (#390)
Wade,
Good find!
You have correctly identified a possibly-old and subtle bug. I imagine the dist call was originally intended to be a data class coercion, not a call to calculate euclidean distance. Wow that's a bad one!
I have updated the github devel version, link should post here any second...
Thanks for the great catch! Always very appreciated, especially just-before a big BioC snapshot update [:+1:]
— Reply to this email directly or view it on GitHubhttps://github.com/joey711/phyloseq/issues/390#issuecomment-58307241.
I recently ran into a problem using the distance function with the method “dpcoa” where the results were not the same as what came from the dpcoa function in ade4. I examined the code and found where the difference is coming from. Below is what I found:
From above, I see that ‘DPCoA()’ is wrapped with ‘dist()’. Unless I am mistaken, this would cause R to calculate the Euclidean distance of the DPCoA distance matrix. Is this what you intended?