Dennis,
Dennis,
Voici 2 exemples de calcul des 2 types de scores
Cécile

Cas des singletons : score de prediction d'une realite biologique

Dans le fichier FoldChange0.csv, la CIBLE est :
- 50 fois a  0
- 71 fois a  1
-  3 fois a -1

Alternative predicting genes: ['FT1'] for target: CIBLE with percentage correct: 0.81
  FT1 = 1 correctly predicts 0 x1 times out of y1 with prob correct: p1 and p-value pval1
  FT1 = 0 correctly predicts 0 x2 times out of y2 with prob correct: p2 and p-value pval2
  FT1 = -1 correctly predicts 1 x3 times out of y3 with prob correct: p3 and p-value pval3

Calcul du score : utilisation des lignes ou (FT1 et CIBLE sont à 0) et (FT1 et CIBLE sont à non0)
s = (x2 + x3) / (50 + 71)
ce score doit etre superieur a 70% (variable) pour que le facteur de transcription soit sélectionne pour la suite

===========================================================================

Cas des paires, triplets, quadruplets : score de
classement

Alternative predicting genes: ['FT1', 'FT2', 'FT3'] for target: CIBLE with percentage correct: 0.88
  FT1 = 1 correctly predicts 1 x1 times out of y1 with prob correct: p1 and p-value pval1
    FT3 = 1 correctly predicts 0 x1.1 times out of y1.1 with prob correct: p1.1 and p-value pval1.1
    FT3 = 0 correctly predicts 1 x1.2 times out of y1.2 with prob correct: p1.2 and p-value pval1.2
    FT3 = -1 correctly predicts 1 x1.3 times out of y1.3 with prob correct: p1.3 and p-value pval1.3
      FT2 = 0 correctly predicts 1 x1.3.1 times out of y1.3.1 with prob correct: p1.3.1 and p-value pval1.3.1
      FT2 = -1 correctly predicts 1 x1.3.2 times out of y1.3.2 with prob correct: p1.3.2 and p-value pval1.3.2
  FT1 = 0 correctly predicts 0 x2 times out of y2 with prob correct: p2 and p-value pval2
    FT3 = 1 correctly predicts -1 x2.1 times out of y2.1 with prob correct: p2.1 and p-value pval2.1
      FT2 = 1 correctly predicts -1 x2.1.1 times out of y2.1.1 with prob correct: p2.1.1 and p-value pval2.1.1
      FT2 = 0 correctly predicts 0 x2.1.2 times out of y2.1.2 with prob correct: p2.1.2 and p-value pval2.1.2
    FT3 = 0 correctly predicts 0 x2.2 times out of y2.2 with prob correct: p2.2 and p-value pval2.2
      FT2 = 0 correctly predicts 0 x2.2.1 times out of y2.2.1 with prob correct: p2.2.1 and p-value pval2.2.1
      FT2 = -1 correctly predicts 0 x2.2.2 times out of y2.2.2 with prob correct: p2.2.2 and p-value pval2.2.2
    FT3 = -1 correctly predicts 0 x2.3 times out of y2.3 with prob correct: p2.3 and p-value pval2.3
      FT2 = 1 correctly predicts 0 x2.3.1 times out of y2.3.1 with prob correct: p2.3.1 and p-value pval2.3.1
      FT2 = 0 correctly predicts 0 x2.3.2 times out of y2.3.2 with prob correct: p2.3.2 and p-value pval2.3.2
      FT2 = -1 correctly predicts 0 x2.3.3 times out of y2.3.3 with prob correct: p2.3.3 and p-value pval2.3.3
  FT1 = -1 correctly predicts -1 x3 times out of y3 with prob correct: p3 and p-value pval3
    FT3 = 0 correctly predicts -1 x3.1 times out of y3.1 with prob correct: p3.1 and p-value pval3.1
    FT3 = -1 correctly predicts 0 x3.2 times out of y3.2 with prob correct: p3.2 and p-value pval3.2

Calcul du score : utilisation des lignes ou (FT1 et CIBLE sont à 0) et (FTn et CIBLE sont à 0) et (FT1 et CIBLE sont à 1/-1) et (FTn et CIBLE sont à 1/-1)
si tous les intermediaires entre 1 et n ont le meme profil (0/0 ou non0/non0) :
score de classement : s2 = (x1.3.2 + x2.2.1) / (x1 + x2)