Bacterial GWAS

Overview

Teaching: 10 min
Exercises: 30 min

Questions

• Which genes are associated with patient mortality

Objectives

• Determine genes significantly associated with patient mortality and speculate why

Introduction

As mentioned in the first part of this exercise series, the 62 S. pneumoniae isolates were from patients who suffered from a invasive pneumococcal disease (IPD). IPD, especially in elderly is associated with high mortality. The outcome is partially explained by co-morbidities (e.g. diabetes) but there are some indications that it matters which strain of S. pneumoniae is causing the infection.

We will therefore try to find out if there are specific virulence factors that are associated with patient survival.

Bacterial Genome Wide Association Studies (GWAS)

An excellent primer on bacterial GWAS is available here: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3743258/ . We will make use of the tool “Scoary” ( https://github.com/AdmiralenOla/Scoary ), a tool complementary to Roary. Scoary has implemented several GWAS methods into one tool. Scoary needs a comma separated list of strain characteristics (phenotypes).

Copy and paste the columns with the names and patient outcome in a file. Any texteditor can be used but in this example mcedit is used. The file can also be made on your own computer and uploaded using filezilla, scp, etc.

$ cd ~/orthology
$ mcedit traits.txt
#copy paste the two columns, save the file, exit mcedit
$ cat traits.txt |tr "\t" "," > traits.csv

We can now combine our gene presence absence table with the patient data and check if there are indeed specific virulence factors associated with patient mortality. We’re using the special keyword “ALL” the include all annotation data. The P-value cutoff is 0.01 to limit the output

$ cd ~/orthology
$ scoary -t traits.csv -g gene_presence_absence.csv  -p 0.01 --include_input_columns ALL

Discussion: Which genes are associated with patient mortality. Where are they located? what is their function?

Download the file mortality_.results.csv and open in Google Spreadsheet or Excel or any other spreadsheet tool. Investigate the outcome, look at the column with OXC141 as this is a completed reference genome and the order of the genes is known for all genes. Is there something special about the order and location of the genes? It is also possible to open the OXC141 genbank file ( /home/student100/annotation/OXC141/OXC141.gbk ) in IGV and view the best score gene(s) there. Also: Remember the figtree visualization of the tree and the coloring of the phenotypes? Generate a similar annotation file with the locus you think is involved in patient mortality and display it in figtree. Does a specific clone carry the virulence factor(s)?

Key Points

• Contigency testing for gene presence absence to associate a genotype with a phenotype, similar to GWAS in clinical genetics is possible with bacterial genomes