analysis/Hsu.et.al.git/prepare.DeepSequence.from.EVE.R

eve.folder <- '/share/vault/Users/gz2294/EVE/results/evol_indices/'
genes <- c('PTEN', 'NUDT15', 'CCR5', 'CXCR4', 'GCK', 'CYP2C9', 'ASPA', 'SNCA')
for (gene in genes) {
  # read EVE all scores
  eve <- read.csv(paste0(eve.folder, gene, '_HUMAN_20000_samples.csv'))
  for (fold in 0:4) {
    data.gene <- read.csv(paste0('data/', gene, '.fold.', fold, '.score.1/data.csv'))
    data.gene.eve <- data.frame(mutant=data.gene$mutant,
                                mutation_effect_prediction_vae_ensemble=eve$evol_indices[match(data.gene$mutant, eve$mutations)])
    print(sum(is.na(data.gene.eve)))
    dir.create(paste0('inference/', gene, '.fold.', fold, '.score.1/DeepSequence/'), recursive = T)
    write.csv(data.gene.eve, paste0('inference/', gene, '.fold.', fold, 
                                    '.score.1/DeepSequence/vae_predictions.csv'))
    for (subset in c(1, 2, 4, 6)) {
      data.gene <- read.csv(paste0('data/', gene, '.subset.', subset, '.fold.', fold, '.score.1/data.csv'))
      data.gene.eve <- data.frame(mutant=data.gene$mutant,
                                  mutation_effect_prediction_vae_ensemble=eve$evol_indices[match(data.gene$mutant, eve$mutations)])
      # print(sum(is.na(data.gene.eve)))
      dir.create(paste0('inference/', gene, '.subset.', subset, '.fold.', fold, '.score.1/DeepSequence/'), recursive = T)
      write.csv(data.gene.eve, paste0('inference/', gene, '.subset.', subset, '.fold.', fold, 
                                      '.score.1/DeepSequence/vae_predictions.csv'))
    }
  }
}