module04 feedback

[tavareshugo]

Attendance: 14

Ideas for improvement of these materials

Using copy-paste from materials is a bit clunky. Can we use inbuilt datasets? Or start by setting working directory and read data from data directory

For example, we could have two files:

• a "wide" dataset of expression data genes x samples - ensure samples have some "metadata" in their name, e.g. "wt_1", "wt_2", "mut_1", "mut_2"
• a gene annotation table

Then we want to:

1. gather our data
2. join data with annotation
3. separate the "sample" column into genotype and replicate
4. find the top expressed genes (arrange)

Also:

• for gather and spread:
  • have more than 2 columns to gather to show example of colX:colY
• for join:
  • show example of joining by more than one variable
  • show example of joining columns of different names
• for arrange:
  • show example of sorting by more than one variable
  • show example of sorting character column
• for factor:
  • mention can also be used to change order of legend

Exercise solutions:

• Maybe in the example of geom_violin can use a log-scale to illustrate that that would be a better visualisation of the data.

[tavareshugo]

Feedback:

• Not enough time for exercise
• Have breaks?