User:Steven J. Koch/Notebook/Kochlab/2010/04/30/Tracking 18% 18-oxygen water results
Steve Koch 23:34, 30 April 2010 (EDT): As a combination of curiosity and wanting to get accustomed with the tracking, I used manual tracking for a couple hours to look at Andy's 18% heavy-oxygen data today. I was expecting it to look "normal." And it might. But the analysis (see below for evernotes) showed disting two populations of speeds. I am very curious whether these are the two dominant protofilament numbers? There is also a second (or perhaps two) peak at much lower speed. Maybe another population of protofilament number?
- Steve Koch 23:40, 30 April 2010 (EDT): Snapshots of analyses can be found here: http://kochlab.org/files/data/Microtubule%20Tracking%20Data/2010%2004%2030
Steve Koch 23:27, 1 May 2010 (EDT): I realized that Larry's kernel density estimator wasn't working because I had an NaN in the array. So, I put it back in. It didn't look any different than my sliding window, except it is normalized. However, his "auto h" method suggests 19 for the bandwidth, instead of the 10 I was using. When it's 19, you can't see the two peaks. But, if I exclude the two points below 600, it suggests a bandwidth of 14 and you can see the two peaks. Thus, I am now thinking we need to set the bandwidth using simulated data with known spreads. Links:
- http://www.evernote.com/pub/skoch3/Test#45102ff1-da0f-4aa9-90e2-464a57e403bd
- http://www.evernote.com/pub/skoch3/Test#a0aafab3-fc13-430e-85c2-d2bcc973b8c2
- http://www.evernote.com/pub/skoch3/Test#08add350-c274-4cc8-b854-763f65676043
Evernotes
<html><iframe width="1000" height="600" frameborder="0" src="http://www.evernote.com/pub/skoch3/Test#n=b4fb8233-0fa3-404a-91c2-997d3bb2295f" ></iframe></html>
