IMPACT Checklist Vignette

[karissawhiting]

Create a vignette in {gnomeR} to guide people through IMPACT data checks with explanations and examples

• [ ] Decide what data set /samples to use.
• [ ] Create branch and Rmd document for Esther to work on during hackathon
• [ ] Karissa will add example code for the first item below using cbioportalR once we decide on example samples

Pre Analysis Sample Checks

• Pull clinical and genomic and compare the two to make sure all were sampled. Get list of samples to make sure there are none missing or incorrect (we can use get_clinical_by_sample()). Find an example of this or create one where the investigator gives you a sample with incorrect ID?
  • In {cbioportalR} - Maybe add a more descriptive warning message of which samples were not found.
• Duplicate patient check when multiple samples per patient - (helper function?)
• Check cancer type against Oncotree code (helper function?)(maybe add oncotree dictionary?)
• Also check sample_type (primary/mets) against clinical data. (helper function?)

Genomic data Checks

• Gene alias check (just write a note about why we need to do this and how gnomeR handles this)
• Data checks between research and IMPACT samples. Send an example to Esther with current messaging to see if she things it's descriptive enough
• Any germline? remove them. Also there are blanks NA. These are likely somatic without matched pairs. Suggest that ppl do sensitivity analysis.
• Compare TMB between research samples & IMPACT
• Something about threshold - maybe if small n do x% in either group?
• oncoKB cancer type - maybe add note about how you need this for oncoKB annotating ?

Maybe if you have research samples those related items are separated into a different section?