QTL-analysis

Script(s) for standard QTL analyses. QTL analyses use penalties calculated from one-,two- dimensional scans/permutations for single-, multiple- QTL mapping via R/qtl. This includes bash scripts to calculate permutations on the Texas A&M Institute for Genome Science & Society High Performance Cluster (tigss-hpc).

Input data:

The single and multiple QTL mapping R scripts (single_qtl_mapping_rqtl.R, multiple_qtl_mapping_rqtl.R) both use as input an R/qtl cross object. The input R/qtl cross object is constructed and saved with save_crossobject_rqtl.R. The specific cross used in the script can be found at the MulletLab GitHub which is part of the Supplemental Information from Truong, McCormick, Rooney & Mullet (Genetics, 2015):

• R07018xR07020_genetic_map_with_phenotypes.csv

The scripts used to calculate penalties for multiple QTL model traversal (as described by Manichaikul et al. (Genetics, 2009)) is also found on the MulletLab GitHub.

Requirements:

These scripts (save_crossobject_rqtl.R, single_qtl_mapping_rqtl.R, multiple_qtl_mapping_rqtl.R) require the R/qtl library developed by Professor Karl Broman, and are not tied to any architecture.

Permutation bash scripts found in the tigss-hpc_rqtl_penalty_folders.zip are specific to the tigss-hpc architecture, however the workflow can be modified to be of use elsewhere.

Typical workflow:

1. Construct cross object with single_qtl_mapping_rqtl.R
2. Calculate penalites for QTL mapping in compute cluster (tigss-hpc, see configuration_tigss-hpc_steps.txt to set up)
3. Modify parameters and filenames in folders:
   • /data/{username}/rqtl_crosses/
   • /data/{username}/rqtl_mqm_scripts/
4. Submit jobs ($ ./data/rqtl_mqm_scripts/run_two-dimensional_scans_job_array_{crossname}.sh)
5. Combine penalties ($ ./data/rqtl_mqm_scripts/run_combine_scans_{crossname}.sh)
6. Get penalty scores from ($ vim /data/rqtl_mqm_output/{crossname}/mqm_scantwo_penalties/0p05_penalities.txt)
7. Run single-QTL mapping with single_qtl_mapping_rqtl.R with

• crossobject from step 1
• main penalties from step 2

4. Run multiple-QTL mapping with multiple_qtl_mapping_rqtl.R with

• crossobject from step 1
• main penalties from step 2
• initial model with main QTL identified from step 3