bas                   package:OLIN                   R Documentation

_B_e_t_w_e_e_n-_a_r_r_a_y _s_c_a_l_i_n_g

_D_e_s_c_r_i_p_t_i_o_n:

     This function performs an between-array scaling

_U_s_a_g_e:

     bas(obj,mode="var")

_A_r_g_u_m_e_n_t_s:

     obj: object of "marrayNorm"

    mode: mode of scaling. Default option is scaling of arrays   to
          have the same within-array variance of  logged ratios
          ('var'). Alternatively, 'mad' 'qq' can be used (see details)

_D_e_t_a_i_l_s:

     The function 'bsv' adjust the scale of  logged ratios
     ('M=(log2(Ch2)-log2(Ch1))')  between the different arrays stored
     in 'obj'.

     Following schemes ('mode') are implemented:

        *  'mode="var"':  Logged ratios 'M' are scaled to show the same
           (within-array)  variance for all arrays in the batch stored
           in 'obj'.  The variance is calculated using 'var'.

        *  'mode="mad"':   The same procedure as for 'mode="var"' is
           applied using, however, median absolute deviation ('mad') as
           robust estimate for withing-array variance.

        *  'mode="qq"':  The _quantile scaling_ is using the same
           procedure as the quantile normalisation described by Bolstad
           et al. (2003). In brief: Given X is the matrix with logged
           ratios (column corresponding to arrays, rows to genes) 

           1.  Sort each column  of X (independently) producing Xs, 

           2.  Replace values in each row of Xs  by the mean value of
              the row producing Xsm,

           3.  Rearrange the ordering  for each column of  matrix Xsm, 
              so that it has the columns have same ordering as for the 
              original matrix X. 

_N_o_t_e:

     Between-array scaling should only be performed if it can be
     assumed that the different arrays have a similar  distribution of
     logged ratios. This has to be check on a case-by-case basis.
     Caution should be taken in the interpretation of results for
     arrays hybridised with  biologically divergent samples,  if
     between-array  scaling is applied.

_A_u_t_h_o_r(_s):

     Matthias E. Futschik  (<URL:
     http://itb.biologie.hu-berlin.de/~futschik>)

_R_e_f_e_r_e_n_c_e_s:

     Bolstad et al., A comparison of normalization methods for high
     density oligonucleotide array data based on variance and bias,
     _Bioinformatics_, 19: 185-193, 2003

_S_e_e _A_l_s_o:

     'marrayNorm','var','mad'

_E_x_a_m_p_l_e_s:

     # DISTRIBUTION OF M BEFORE SCALING
     data(sw.olin)

     col <- c("red","blue","green","orange")
     M <- maM(sw.olin)

     plot(density(M[,4]),col=col[4],xlim=c(-2,2))
     for (i in 1:3){
       lines(density(M[,i]),col=col[i])
     }

     # SCALING AND VISUALISATION  
     sw.olin.s <- bas(sw.olin,mode="var")
       
     M <- maM(sw.olin.s)

     plot(density(M[,4]),col=col[4],xlim=c(-2,2))
     for (i in 1:3){
       lines(density(M[,i]),col=col[i])
     }

