Hi, I asked people at Affymetrix and when they write in their Statistical Algorithms Description Document about background correction "The cells are ranked and the lowest 2% is chosen as the background b for that zone (bZk)" , they mean to take the average of the lowest 2% for the background value for each section. Therefore, a modification like this of the file chipbackground.c (adding the line bg_q[j] = sumx;) gives signal values more in agreement with the MAS 5.0 software. Hélčne /********************************************************************* * **for (j=0; j < grid_dim; j++){ ** bg_q[j] = data_by_sector[j][(int)(0.02* nprobes_in_sec[j])]; **} ********************************************************************** */ for (j=0; j < grid_dim; j++){ sumx = 0.0; sumx2 = 0.0; lower2pc = (int)(0.02* nprobes_in_sec[j]); i = 0; while (i < lower2pc){ sumx += data_by_sector[j][i]; i++; } sumx = sumx/lower2pc; i =0; while (i < lower2pc){ sumx2 += (data_by_sector[j][i] - sumx)*(data_by_sector[j][i]-sumx); i++; } bg_q[j] = sumx; noise_q[j] = sqrt(sumx2/(lower2pc -1)); } free(nprobes_in_sec); free(cur_n); free(data_by_sector); }