Although we may start discussing migration tomorrow, I'm unlikely to have revised notes posted before lecture. (You can look at the old ones by following the link from the "Notes" page, rather than the non-existent link from the lecture detail page.) I apologize for not having them ready, but I'm going to be spending the rest of my afternoon and evening grading papers from Project #1. I want to hand them back and to talk a little about the results tomorrow.
February 2004 Archives
1. Remember: Think Backwards! Or at least work backwards. If you're having trouble getting started writing your WinBUGS code, apply one of the seven habits of highly effective people by beginning with the end in mind. Start by thinking about what you want your WinBUGS output to include, i.e., what you want WinBUGS to calculate for you, to answer each of the questions. Then see if you can figure out how to get WinBUGS to give you numbers for those things.
2. The difference makes no difference. Whether it's positive or negative, that is. If you need to compare two values, and you want to do that by subtracting one from the other, it doesn't really matter which one you subtract from the other. All you really want to know is whether the difference between them is different from zero, so it doesn't matter whether that difference is positive or negative.
Does this mean I get to spend spring break GRADING???
Given the revised due date for Problem 2, I would like to reschedule my "help" times. I will plan to be available in the TLS computer room Mon. after class rather than today (maybe by Mon. I'll have figured this stuff out). If you have specific questions now I will be happy to talk to you after class today. Beach Hall folks, is there any time Tues. that most of you could gather 'round the computer? My schedule is pretty flexible that day, just let me know.
Feeling your pain,
Robynn
I've had enough questions and reports of difficulties with problem #2 that I'm going to switch the schedule of assignments again.
- Problem #2 will be due on Friday, 5 March instead of next Monday.
- I'll assign Problem #3 on Wednesday, 3 March instead of next Monday
- Problem #3 will be due on Monday, 22 March instead of Friday, 19 March
I'll announce the change in class this morning, but the changes are reflected in the on-line syllabus right now.
Just a reminder. I mentioned on the first day of class that North Carolina State University offers a Summer Institute in Statistical Genetics, a series of 2 1/2 day short courses on a variety of topics related to what we cover in this course. Scholarships are availaable. From the web site:
Scholarships
A few tuition and/or travel scholarships are available for undergraduate, graduate, and postdoctoral students wishing to attend the Summer Institute in Raleigh. Applications are due on March 31, 2004, and should include a letter explaining the reason for interest in the Institute, a one-page CV, and a letter of support from the student's academic advisor.
For more information on the modules offered, including one on population genetic data analysis (featuring yours truly) and another on interpreting DNA evidence (for forensic scientists and evolutionary biologists) go to http://statgen.ncsu.edu/brcwebsite/summer_institute_ral.php.
You all know WinBUGS well enough not to need any help with Problem 2, right? Ok, good, well I'll be in the TLS computer room Fri. after class so you can explain it to me. I can stop by Beach Hall Thurs. at 1:00 to get the Strausbaugh lab version of things if that works for those of you who care.
As always, you can e-mail me any time.
Robynn
In writing our my solution to the problem, I realized that I need to change one of the hints. Use this
p.poly <- step(s.1)*step(s.2)*(s.2/(s.1+s.2)) + step(-s.1)*step(s.2)
Instead of what I suggested in the notes.
I've posted notes on selection components analysis to the web site. Since we won't be discussing it in lecture, I haven't linked to them from any individual lecture. You can find them from the "Notes" page. You'll notice that there isn't anything about a Bayesian approach to selection components analysis in the notes. As with the other analyses we've looked at, a Bayesian approach has distinct advantages, but I didn't have time to develop the notes to illustrate how to do selection components analysis in a Bayesian context. If you're interested, ask. The principle is straightforward, even if the implmentation is messy.
I've also posted notes for Wednesday and Friday's lectures on genetic drift. It's the same set of notes for both lectures, so there's no need to print them twice. I've also included a link to a paper you may (or may not) want to look at. It will give you some idea of what's known about drift, migration, and mutation. It is, however, very heavy going, so take a look at it on screen before printing it out. You may decide that you really don't want to read it, which would be just fine. I'm just providing a pointer on the remote chance that you'd actually like to try.
The notes for Monday's lecture are now posted. We'll start by finishing our discussion of how to estimate viabilities. Then we'll talk briefly about selection at one locus with multiple alleles. The notes include information on fertility and sexual selection, but I don't think we'll have time to discuss them. I want to start discussing genetic drift on Wednesday.
As promised, I've posted extensively revised notes on the genetics of natural selection. We'll probably be using for at least part of Friday too. Enjoy!
It's been brought to my attention that some of you are getting inconsistent results from Structure. Specifically, the folks from Beach Hall report that some of them find the following for the "Estimated Ln Prob of Data":
- K=7: -2725.6
- K=8: -3093.4
Others get
- K=7: -2723.3
- K=8: -2698.4
This illustrates a property of MCMC implementations of Bayesian analyses I'd hoped to avoid for awhile. Let me give a brief explanation here, and we'll talk more about it when we return to estimating migration rates (with Structure) in a week or two.
The calculations that Structure does have a random element. If you run the analyses long enough, that random element will disappear. In this case, the close agreement for the K=7 results suggests that a longer run will give the same results. (You could try this and check it to see.) The disagreement between the K=8 results suggests that a longer run will give results different from one or both of the ones you've tried. My bet is that a long run will settle on the one with P(data) = -2698.4.
For purposes of Project #1, you may want to try longer sample runs than what I initially suggested (if you have time), but I won't require that. Just make sure that you include the numerical results that caused you to chose the K you did, so that I can verify that you're doing that part of the analysis correctly.
The introductory notes on natural selection have now been posted. We'll be working from these notes through Wednesday, after we finish up with our discussion of two-locus genetics.
I've also revised the lecture schedule to reflect the recent changes.
Thanks to some prompting from your very conscientious TA, I've taken a second look at the lecture and assignment schedule. As a result, I've decided to make some changes.
Instead of Project #1 being due next Monday, 16 February, I've decided to extend the deadline to Wednesday, 18 February. I've also adjusted the dates associated with Problem #2. The changes are reflected in the version of the syllabus that now appears on the web site.
I'll be adjusting the lecture schedule over the weekend, too. Right now I anticipate spending about half of Monday's lecture on two-locus population genetics and then starting with the genetics of natural selection. That puts us about a lecture and a half behind where I'd planned, and I'm not sure yet where I'll make up the slack. One possibility, unless there is a major outcry, is to drop the lecture on fertility and sexual selection. Let me know soon if you wuold rather have me drop something else.
If you use Internet Explorer to download the data files for Project #1, they're likely to land on your hard disk with a .txt extension. (Depending on how you have your options set in Explorer, you may or may not see that extension.) You'll need to navigate to the directory where you downloaded the files and rename them manually to have a .nex extension before you'll be able to load them in Hickory (or you could switch to Mozilla).
Here's additional information you'll need to run Structure on the data that's provided.
platanthera-rapd-7.stru
- Number of individualss: 155
- Ploidy of data: 1
- Number of loci: 69
- Missing data value: -9
- Select "Data files stores data for individuals in a single line" in step 3.
- Select "Putative population origin for each individual" in step 4.
platanthera-rapd-14.stru
- Number of individualss: 198
- Ploidy of data: 1
- Number of loci: 69
- Missing data value: -9
- Select "Data files stores data for individuals in a single line" in step 3.
- Select "Putative population origin for each individual" in step 4.
platanthera-allozyme-7.stru
- Number of individualss: 152
- Ploidy of data: 2
- Number of loci: 12
- Missing data value: -9
- Select "Row of marker names" in step 3
- Select "Data files stores data for individuals in a single line" in step 3.
- Select "Putative population origin for each individual" in step 4.
That should give you everything you need to know for the Structure analyses. E-mail me if you have any questions.
The notes on two-locus population genetics have now been updated. Although the plan was originally to finish that topic tomorrow and start with the theory of natural selection on Friday, we probably won't make it. I'd like to take some time at the beginning of class tomorrow to re-visit individual assignment procedures using Structure and then take some time for questions about analysis of population genetic structure before moving on. Please come prepared with questions.
The notes for today's lecture on assignment are now available. Sorry for the delay. I'll also review use of Structure at the beginning of lecture on Wednesday. I clicked a lot of buttons pretty quickly today, and I want to go back and do it more slowly.
The notes on Bayesian F-statistica are available. We probably won't finish talking about the Bayesian approach today, but we should at least get started.
Don't forget that in addition to the first problem being due today, I'll be assigning the first longer project on Monday.
A couple people have stopped by to ask questions about the problem set today. To make sure that everyone knows, I am available for questions (if you can find me), and you're free to ask any question you want about the problem set. But I may or may not answer the question, and if I answer the question I may do so only partially. I don't want to give everyting away.
When I was coding my own solution to Problem #1 this morning, I found a couple of typos in the hints. They're places where (a) the suggested code doesn't match the formulas in the table (the table is right) and (b) a prior distribution that is missing (I suggest a uniform prior on f).
The corrected version (in both HTML and PDF) is available on the page for the 30 January lecture. There's no need to print the new version out, unless you want to. You can simply copy the corrections to your code.
Sorry for the confusion.
I haven't run this in WinBUGS yet, but the following should work to get the data iin. Let me know if you run into any more problems.
list(k.ff=267, n.ff=329, k.fs=c(81, 120, 44), n.fs=245, k.ss=143, n.ss=313)
The version of the problem on the web site now reflects this change.
I will be in the TLS computer room (TLS 477) Wed. from 2:00 to 4:00. I would be happy to do a separate session in MCB as well, if you want and if you have a few computers together that have WinBUGS on them. I could come over there any time Thurs.
Remember, you can e-mail (or try calling or stopping by) any time, especially if Wed. afternoon is not good for you.
Robynn
The notes for Monday, 2 February are now posted.
I've included a link in the lecture description to updated notes on identity by descent. Because we had to cancel lecture last Wednesday, I'm not going to have time to cover that material in lecture. You may want to look at it on your own, although I don't think that any of the problems or projects this semester will require that you understand the concept.
