QBB2024 - Day 1 - Lunch Exercises

Overview

Biological Learning Objectives

• Explore the Adult GTEx project sample metadata
• Describe the results of your data analysis

Computational Learning Objectives

• Explore text files using R and dplyr
• Document your work on GitHub

Instructions

Document your answers in a single file stored in ~/qbb2024-answers/day1-lunch.

Please git push after each exercise and do not wait until the end of the session e.g.

git add explore-samples.Rmd
git commit -m "Add answer for exercise 1"
git push
# Confirm at github.com

Exercises

1. Browse data dictionary

   • Navigate using a web browser to https://gtexportal.org/home/downloads/adult-gtex/overview
   • Select data type: Metadata
   • Download and take a quick look at GTEx_Analysis_v8_Annotations_SampleAttributesDD.xlsx

2. Prepare your working environment

   • Reset R by opening the Session menu and selecting “Terminate R…”
   • Create either a new R Script (e.g. explore-samples.R) or a new R Notebook (e.g. explore-samples.Rmd) using the provided template
   • Load the tidyverse package

3. Wrangle the sample metadata

   • Load the GTEx_Analysis_v8_Annotations_SampleAttributesDS.txt file and assign to variable df

   • Create a SUBJECT column using the following code

       df <- df %>%
         mutate( SUBJECT=str_extract( SAMPID, "[^-]+-[^-]+" ), .before=1 )
     

   • Confirm that the first column in df is SUBJECT

4. Which two SUBJECTs have the most samples? The least?

   • Use group_by(), summarize(), and arrange()
   • Describe your results as either a comment (# in R Script) or a bullet point (- in R Notebook)

5. Which two SMTSDs (tissue types) have the most samples? The least? Why?

   • Solve and document as above

6. For subject GTEX-NPJ8

   • Filter for samples from this subject and save as a new object (e.g. df_npj8)
   • Which tissue has the most samples?
   • For that tissue, what is different between the samples? Scroll to the 15th through 20th columns …

7. Explore SMATSSCR (autolysis score)

   • Filter out NA values in this column to avoid mean() returning NA

       df %>%
           filter( !is.na(SMATSSCR) )
     

   • How many SUBJECTs have a mean SMATSSCR score of 0?
   • What other observations can you make about the distribution of mean scores?
   • What are possible ways to present this information in a report?

Just for fun

A. Identify another handful of columns using the data dictionary, explore the data, and describe your results