Compare NTR, MET, YFS, and Predixcan

[kunalbhutani]

Compare the differences in weights learned by the different methods for two studies

• How much do the beta's for a target snp differ?
  -- Relative to a gene's overall expression?
  -- Does this position have high LD with another position? Colinearity?
• Quantify the differences in predicted expression
  -- Does using each method separately still control type I error?
• How do their predictions differ from GTEX Blup estimates?
  -- Should you be matching on genetic background initially?
  -- Is one method consistently better for all genes?