John's Blog

"Macro" vs. "micro" evolution

So about 6 years ago was the first time I heard a fundamentalist Christian make a distinction between "macro-evolution" and "micro-evolution." The former (supposedly not observed) refers to speciation over long periods of time whereas the later refers to adaptation that can be seen over shorter periods in the lab, like antibiotic resistance for example. Anyway, this week on fark of all places, somebody posted the the best simple rebuttal to this arbitrary (and artificial) distinction I have read.

Macro-evolution is just micro-evolution over a really long time. Most people who say this kind of thing have no grasp on how long billions of years actually is.It's like me saying that walking from one end of my living room to the other is "micro-walking", and walking from New York to Los Angeles is "macro-walking". The only difference is related to scale and time; the process itself is exactly the same. Biological evolution is change in the gene pool of a population over time. That's all it is.

So anyway, micro-walking now joins intelligent falling as one of my favorite concepts.

John E. Hayes, PhD

So at risk of sounding like a naive high schooler quoting the Dead, let me say what a long strange trip it's been, but I finally made it. The odd part is that I had essentially the same reaction as I had when we got married - awaking up the next morning, nothing really felt different. I mean, it was nice to make everything official, but it wasn't some dramatic seachange.

My defense went very well or so they tell me. :) It was all a bit of a blur, but I felt good up at the front of the room, like I was totally on top of my game and in the place I needed to be to really nail it. I must have done a bang up job on the dissertation as well, because my committee didn't even want any corrections and said I could file it with the grad school as is. I thought everybody had to do at least some corrections. Anyway, one chapter is published already, only three to go. Then Val and I only have another 37 projects to wrap up. :)

It's done.

My dissertation draft is done.

http://duffylab.ah.uconn.edu/pubs/hayes/

Now I just need to sleep on it and have Val notify the graduate faculty.

Then on Monday I get to file some last minute paperwork and start making slides. Fun Fun.

Happy Earth Day: Change your WinXP settings

So it turns out that the standard Windows standby mode (known as standby S1) only saves about 10% in energy consumption. Microsoft has a second more efficient standby mode (standby to ram, known as S3 standby) that powers down everything except the ram. It uses a small fraction of full power with still essentially being instant on.

Here's how to enable it:

http://www.exoid.com/?p=29

http://www.yafla.com/dennisforbes/On-PC-Power-Saving-S1-to-S3-standby-dumppo-exe/On-PC-Power-Saving-S1-to-S3-standby-dumppo-exe.html

good wine in a box

Slate had this interesting article http://www.slate.com/id/2146720 which eventually lead me to this webpage:
http://www.dtourwine.com/index.html

The idea of good French wine in a container where it won't oxidize really appeals to me. You could keep one of these in your fridge and have a glass or two of wine with dinner for two weeks straight without having to worry about the wine going bad. All for a price per glass comparable to a decent bottle of beer. Too bad I live in Rhode Island that doesn't allow internet wine sales.

18 1/2 missing emails

So I was talking with Beeeej tonight about the thousands of missing emails that the White House claims it can't find. Sen. Leahy doesn't believe they are really missing, because, in his words "You can't erase e-mails, not today. They've gone through too many servers."

Then it hit me. I wonder if NSA has a copy of the emails stashed away in Ft Meade as part of the automated warrantless wiretapping that resulted from the Patriot Act?

Now that would be ironic.

ROC Curves - Theory, Explanation and SAS

So I was thinking about putting some ROC curves into my dissertation but I decided I don't have the time to learn something new right now, so I'm bookmarking these pages for later.

Here's a great explanation:
http://www.anaesthetist.com/mnm/stats/roc/Findex.htm

And another, less interactive page:
http://www.childrensmercy.org/stats/ask/roc.asp

And here are two pages on making ROC Curves in SAS:
http://support.sas.com/faq/031/FAQ03116.html
http://www.umanitoba.ca/centres/mchp/concept/dict/Statistics/ROC.html

SAS/ODBC bug when importing from MySQL

So I store a bunch of my dissertation data in a MySQL database running on OSX, storing the different pieces of my data in separate tables, with common user IDs across the tables. Then when I'm in SAS on my windows box, it is trivial to connect to the database with a single line of code.

libname tastedb ODBC datasrc="tastedata" user=XXX password=XXX

Reading in the separate tables from MySQL into unique datasets in SAS is similarly trivial, thanks to the ODBC widget invoked above.

data gene;
set tastedb.genedata;
KEEP id genotype;
run;

data phenotype;
set tastedb.phenotypedata;
KEEP id variable_x variable_y variable_z;
run;

This makes it wicked easy to merge large amounts of data using SQL left joins right in SAS. Here's the code:

PROC SQL FEEDBACK;
CREATE TABLE merged AS
SELECT *, gene.ID as geneID, phenotype.ID as phenotypeID
FROM gene LEFT JOIN phenotype
ON gene.ID=phenotype.ID;
run;
quit;

So once you store your data in MySQL, retrieving lots of data is very easy. More importantly, a little legwork upfront saves you countless hours that would otherwise be spent cutting and pasting in excel to merge datasets for analysis. Now, if you only had a few variables, it's not so bad, but our food frequency questionnaire is hundreds of variables long just by itself.

But there is one little caveat - SAS appears to have a bug where it can get cranky depending on the field type in MySQL. I just spend 45 minutes chasing down an incredibly vague error message in SAS

"ERROR: Invalid sequence of commands for file TASTEDB.foo.DATA"

It turns out that the import from MySQL will fail if you have a text field where all the values are NULL. In one of my tables I had two text fields, Notes1 and Notes2, that were currently empty, in case I wanted to make any comments about that row (subject) in the future. Apparently, SAS doesn't like this. The solution was to add a single comment to the first field and to delete the second.

firefox user agent

Ok, this is really a note to myself, but it may help somebody else.

Anyway, WebCT Vista didn't support firefox 1.5 last year, so I had to spoof the user agent string to bypass the stupid browser detection. Here's how I did it:

1) enter about:config as an address in the address bar of FireFox
2) right click to get the context menu and select "String" from the menu entry "New". Enter the preference name "general.useragent.override", without the quotes.
3) here is the string I used to spoof Firefox 1.0:

Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.7.5) Gecko/20041107 Firefox/1.0

Anyway, I want to delete this override because I need a *newer* useragent string to use zillow.com.

*sigh* why do stupid web developers still use browser detection scripts in 2007?

More data visualization links

http://babynamewizard.com/namevoyager/lnv0105.html

http://www.readwriteweb.com/archives/ibm_many_eyes_after_one_month.php

http://tools.google.com/gapminder/

http://www.swivel.com/