September 28, 2006

Problem #2 posted

I just posted Problem #2. We'll start Friday's lecture by discussing it. If you have a chance, you might want to take a look at it before then. (Of course, I didn't give you much chance to take a look at it.)

Posted by Kent at 8:47 PM | Comments (0) | TrackBack

September 27, 2006

Notes on genetic drift

I've posted the notes on genetic drift. I expect to use this set of notes as the basis for discussion this Friday and next Monday and Wednesday. Genetic drift is a very interesting, important, and subtle phenomenon.

Posted by Kent at 10:54 AM | Comments (0) | TrackBack

September 25, 2006

Final notes on selection

I just posted the last set of notes on selection. Believe it or not, we're now back on schedule. We will start our discussion of the dynamics of genetic drift on Friday.

On a related note, I've checked everything I can think to check on the server that governs access to notes on the site, and I just finished checking that I could get to them from home (outside the UConn.Edu domain). If you're still having trouble getting to the notes (HTML or PDF), please let me know (a) whether you're using a Windoze machine or a Mac, (b) where you're located when you're trying to get access (and your IP address, if you can find it), (c) what time it was when you were having trouble, and (d) the error messages you receive. I apologize for the difficulties some of you are having, but I can't reproduce the problem and I can't find anything that would cause it.

Posted by Kent at 9:18 PM | Comments (0) | TrackBack

Estimating viability selection

I've just posted notes on estimating viability selection. You'll find them under the detail page associated with today's lecture. I think we'll finish the genetics of natural selection today and at least get started on estimating viability selection. Believe it or not, there's a good chance that we'll be caught up by the end of lecture on Wednesday, so that we can start our discussion of genetic drift on Friday.

Posted by Kent at 8:17 AM | Comments (0) | TrackBack

September 21, 2006

Follow up on Arlequin3

As I suspected, Arlequin3 is only available for Windoze. I've just started to play with it, and the interface looks nicer than in the old version. Unfortunately, the file format seems to have changed a little. We won't be using Arlequin again for awhile, but I'll play around with the new version in the meantime to see what advantages there might be for those of you who are Windoze users. I'll mention those advantages when we return to Arlequin later, but I suspect we'll only be using the old version in class.

I have added the link to the Resources page, so that you can find the link easily in the future.

Posted by Kent at 8:18 PM | Comments (0) | TrackBack

Arlequin

A new version of Arlequin has just been released. I'm at the wrong computer to update the link on the Resources page, but I'll try to get to it tonight. If you're really interested and really impatient, the link is http://cmpg.unibe.ch/software/arlequin3/. Unfortunately, the new version appears to be Windoze only. Unless there are analyses in the new version that the old one can't do (or doesn't do correctly), we'll use the old version in class. I'll point out differences between the versions when we get to them, and I suspect that those of you with Windoze machines will want the new version, but I'll let you know for sure after I've had a chance to play with it a little more.

Posted by Kent at 3:47 PM | Comments (0) | TrackBack

September 18, 2006

Due date for Project #1

I looked more carefully at problem/project due dates this morning, and I've revised them a bit. Project #1 will be due a week from today, Monday, 25 September, as requested. I've also adjusted the assignment and due dates of the next few problems/projects.

Posted by Kent at 11:38 AM | Comments (0) | TrackBack

September 17, 2006

Natural selection

I've posted notes on the genetics of viability selection for Wednesday's lecture. On Monday I hope not only to cover assignment approaches to analysis of genetic structure, but also to cover everything you need to know about two-locus population genetics.

At least I hope to cover everything you need to know about two-locus population genetics for purposes of this course. There's a lot more that we could talk about, but I don't think it's as important for most of you as the other things I'm planning to spend time on. If you're interested in what more there might be to learn, don't hesitate to ask.

Posted by Kent at 5:52 PM | Comments (0) | TrackBack

September 14, 2006

Project #1

Project #1 has been posted. Please take a look at it before you get to class, if you have a chance. If you happen to get this message before class, you might want to (a) download and install Structure on your computer and (b) bring your computer to class tomorrow so you can follow along as we run an analysis.

There are a few tricks to getting Structure installed and running under Windows that Uzay and I will be able to help you with (after I show Uzay the tricks). I can try to help you if you're running a Mac, but I don't know how much help I'll be. If you're a little strange (like me) and enjoy using Linux, I can help you with that, too.

Posted by Kent at 7:21 PM | Comments (0) | TrackBack

September 8, 2006

Inbreeding and WinBUGS

The notes use data from the ABO blood group system as an example of analysing inbreeding with WinBUGS. Because of the combination of complete dominance and co-dominance at this locus, however, the analysis of MN data that we went over in class will be more useful to you in analysing the data from Problem #1. The WinBUGS code is available in inbreeding.txt. Enjoy!

Posted by Kent at 8:08 AM | Comments (0) | TrackBack

September 5, 2006

Chi-squared tests and Fisher's exact tests

A couple of people have pointed out that the notes I posted over the weekend describe a 3x2 contingency table (3 maternal genotypes, 2 paternal alleles). Explanations of contingency tables typically talk about 2x2 tables, which has caused some confusion.

The principles are the same. In both cases we're asking whether the proportions of outcomes are the same across categories. In this case, we're asking whether the proportion of patern A1 alleles (Est-F) is the same across the three maternal genotype frequencies. Computer packages like R, SAS, SPSS, JMP, Stata, and the like know how to do the more complicated calculation. The difference between a 3x2 test and a 2x2 test is like the difference between an ANOVA and a t-test.

If this isn't clear yet (or if I've made things worse), ASK A QUESTION.

Posted by Kent at 12:04 PM | Comments (0) | TrackBack

Update on lecture schedule and a note on Problem #1

I realized over the weekend that trying to answer the first question on Problem #1 with WinBUGS involves more complicated coding than I've introduced so far. If you're interested in how to do it, I'd be happy to do so, but just stick with the chi-squared or Fisher's exact test for now.

If you've looked for changes in the lecture schedule, you'll notice that they haven't appeared. That's because I've decided to leave things as they are. We'll catch up by

  • Skipping the Bayesian version of the ABO example (for now, we'll come back to it later).
  • Skipping the explicit discussion of identity by descent. We'll encounter the concept several more times anyway, and I'll tell you what you need to know about it then.
  • Spending a little less time on testing for departures from Hardy-Weinberg. I'll focus on the Bayesian approach, since you'll need ideas from it to answer the third question on the problem set.
  • Focusing on the conceptual basis of Wright's F-statistics, and skipping some of the algebra I'd initially planned to cover. (Please wipe that smile off your face.)

With those modifications we should finish up on the 15th with everything you'll need to do Project #1, which will be assigned that day.

We'll go over how to use WinBUGS again in class tomorrow. If you'd like to follow along, bring your computer and download the files

Posted by Kent at 10:38 AM | Comments (0) | TrackBack

September 2, 2006

Chi-squared and Fisher's exact test

Uzay and I have fielded several questions about the Chi-squared test and Fisher's exact test. If you're not familiar with the tests or if you just need a refresher, take a look at the Wikipedia page on contingency tables, and follow the links to Peason's chi-squared test and Fisher's exact test for mor information on how they're calculated.

I've also posted some notes giving instructions (HTML, PDF) on how to do the tests in R. If you're unsure about any of this please try to take a look at the Wikipedia entries and the notes I've posted before Wednesday and ask a question about what you don't understand before we get started.

On a different note. If you've looked at the lecture schedule, you can see that we're running a little behind. I'll be rearranging the lecture schedule a bit later this weekend, so stay tuned.

Posted by Kent at 8:57 PM | Comments (0) | TrackBack

September 1, 2006

Update on WinBUGS

If you're planning to bring your laptop to class this morning so you can try WinBUGS, you'll want to download the following two files

Sorry for the short notice, but it didn't occur to me until this morning that I'd neglected to mention these files. See you in a few hours.

Posted by Kent at 6:04 AM | Comments (0) | TrackBack