| layout | page |
|---|---|
| title | PROJECT |
We will be using a large pig and wild boar dataset that can be found in Dryad. In addition, make sure that you follow the link that Dryad provide to the paper that describes this dataset. The distribution contains two different plink files (.ped and .map). These have been zipped separately. It also contains a useful excel sheet that describes the provenance of the samples (populations and GPS positions). Unzip them and put the files in the same folder. When we use PLINK we will need to convert the .ped file to a .bed file. This conversion will also create some other files that will be useful. These files potentially have all the information needed for your analyses (check the plink1.9 webpage for more details about the files).
The plink command you will need is
.\plink --file JWM_Final --chr 1-18 --make-bed --out allsamples(you are free to call it something other than allsamples). You must ensure that the .ped and .map files are in the same folder.
However... The dataset has a very large number of individuals sampled from different breeds and wild boar populations (look at the paper). To make it more feasible to work with, it is necessary to trim the dataset to a smaller number of populations. An example text file with the population identifiers is available here. This is based on data from the following populations:
FIWB - Finnish wild boar
GRWB - Greek wild boar
HRWB - Croatian wild boar
IBWB - Iberian wild boar
ITWB1 - Italian wild boar
ITWB2 - Italian wild boar
PLWB - Polish wild boar
SBWB - South Balkan wild boar
These have been chosen to reflect wild boar diversity in Europe, with a sufficiently small number of populations that you can still visually distinguish the groups in pcadapt PCA plots. You will find that Cattell's rule suggests working with K=2 or K=6.
This example is provided to get you started. You are welcome to base your project purely on this subset, which may allow you to frame hypotheses about adaptive selection between wild boar populations. I suggest that you start with this anyway, to familiarise yourself. After which you can frame your own hypotheses and create a new subset to allow you test them for your project.
Whatever subset you choose, you will need to use plink to trim down the bed file. For the pop1.txt example you would use:
.\plink --bfile allsamples --keep-fam pop1.txt --make-bed --out finalNote that plink only works with the prefix names for the files, and then creates various suffixes.
An important file is the .fam file. The first column contains the population associated with each of your individuals in the selected data. It is important to use this file, rather than the excel list, for illustrating the populations in your pca plots and analyses. By all means use the excel to choose your populations; however note that the excel file is slightly disordered for some populations and you may end up mislabelling individuals in the plots.
Note that the number of chromosomes of the domestic pig/boar is 18 autosomes, although there is potentially some variation in boars. The paper uses the following reference genome: SScrofa10.2. This has 18 autosomes pairs plus sex chromosomes. The map file also contains unplaced SNPs (chromosome 0), which do not have map positions. The plink command at the beginning chooses only the 18 autosomes. The chip is described here.
Remember the dataset was mapped to this version of the genome assembly: SScrofa10.2
(You can also look at the Ensemble version of SScrofa10.2 if you are interested.)
It is probably best to conduct your project with a view to how you will write it up in a report.
When you write your report, you will need to provide an Introduction, for which the original paper will be helpful. The final paragraph of the Introduction will then outline the aims of your research. The Methods section will provide a brief overview of the methods you used (nothing too technical). The Results will provide an overview of what you found. For example, I expect it will contain a number of figures from a PCAdapt analysis - score plot and Manhattan plot, for example. It may also contain examples of the PCA distribution of alleles for SNPs that appear to be under selection. You can also summarise in the Results the gene descriptions for candidate SNPs in NCBI. The Discussion will then contrast the results with your initial aims and expectations, provide some critique of what you have found, and point to future research.
Back to pcadapt.
Back to first page.