What have I been doing in Tromsø the week that has passed?
I have learned to use a lot of computer programs to analyze my data (the data I made in June that I will base my master project on), how to connect the information I get and also a bit about how to present what I find out. And today I presented my own analysis on my own data for the first time. Hopefully this is what I will do the rest of my life and today was the first! So exciting!
Two pictures from my presentation today:
This is a picture of a bar plot that you get when you do a genetic structure analysis in Structure. What you do is that you put your data set into the program and then you use Markov Chain Monte Carlo (MCMC) and some prior values to find out how many genetic groups there are in your data set. You indicate number of steps in the burn in (100 000) and then the number of steps of the MCMC (400 000) and then how many groups (values of K) the program should try and how many times it should try each K (number of iterations). Each bar is one individual and the colours indicate the frequency of each genetic group in each individual. Along the x-axis my samples are sorted by geography.
This, as you can see, is a map of the Northern hemisphere. This is where you can find my moss, Hylocomium splendens. It is found on the whole of the Northern hemisphere and also on Mt. Kilimanjaro. The circles are just another way of showing the bar plot. I have made one circle for each of my nine populations. Also here the colours indicate the three different genetic groups.
As you can see, there seems to be some genetic patters and that the colours group geographically. My groups were East and West North-America, Africa, North and South Europe and North, South, West and East Russia.
So when I get back to Trondheim I will clean my data set and get rid of all my missing data and bad groupings and then I will do all this (and more) again. The analysis to make the bar plot took 10 hours of computational time and for the data I will present in my project in the end, I will have to at least double number of steps. And I have to run the analysis many times with different priors (prior values, things I already know). This is what I will do for the next three months at least 🙂