BGandNorm         package:Agi4x44PreProcess         R Documentation

_B_a_c_k_g_r_o_u_n_d _C_o_r_r_e_c_t_i_o_n _a_n_d _N_o_r_m_a_l_i_z_a_t_i_o_n _B_e_t_w_e_e_n _A_r_r_a_y_s

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

     For Background correction it uses the 'backgroundCorrect' function
     of 'limma' package ('half','normexp'). For Normalization between
     arrays it uses 'limma' function 'normalizeBetweenArrays'
     ('quantile','vsn').

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

             BGandNorm(RGlist, BGmethod, NORMmethod, foreground, 
             background, offset, makePLOTpre, makePLOTpost)

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

  RGlist: an 'RGList' object 

BGmethod: Method for the BG corection. Possible values are: 
          'none','half','normexp'. See ?backgroundCorrect in limma
          package for details 

NORMmethod: Method for Norm between arrays. Possible values can be:
          'none','quantile',vsn'. See ?normalizeBetweenArrays in limma
          package 

foreground: Foreground Signal to be used for the analysis.  Possible
          values are 'MeanSignal','ProcessedSignal' 

background: Background Signal to be used for the BG correction. The
          values can be: 'BGMedianSignal','BGUsed' 

  offset: numeric value to add to the intensities before log
          transforming. The offset shrunks the log ratios towards zero
          at the lower intensities. See limma user guide for details 

makePLOTpre: density Plots, box plots, MVA plots and RLE plots with the
          raw signal

makePLOTpost: density Plots, box plots, MVA plots and RLE plots with
          the normalized signal 

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

     In order to make direct comparisons of data coming from different
     chips it  is important to remove sources of variation of non
     biological  nature that may exists between arrays. Systematic
     non-biological  differences between chips become relevant in
     several obvious ways  especially during labeling and
     hybridization, and bias the relative  measures on any two chips
     when we want to quantify differences due to different treatment
     between two samples. Normalization is the attempt to compensate
     for systematic technical differences between chips, to see more
     clearly the systematic biological  differences between samples.
     First data are background corrected. We produce a Background
     Subtracted  Signal. The Background Signal Used depends on the AFE
     settings for the type  of background method calculation and the
     settings for spatial detrend. Usually,  the Background Signal Used
     is the sum of the Local Background Signal + the  Spatial
     Detrending Surface Value computed by the AFE software.  For the
     Background correction we use the 'backgroundCorrect' function of
     'limma' package with options <'half','normexp'>      This function
     is designed to produce positive corrected intensities.  First, any
     intensity value lower than 0.5 is reset to be equal to 0.5. 
     Besides, and offset value (normally 50) is used. This offset value
     adds  a constant to the intensity values before log-transforming,
     so that the log  ratios are shrunk towards zero at the lower
     intensities. After background correction, data are normalized
     between arrays using  'limma' function 'normalizeBetweenArrays'
     with options <'quantile','vsn'>

     For foreground signal,the user can choose between the 'MeanSignal'
     and the 'ProcessedSignal' and between the 'BGMedianSignal' and the
     'BGUsed' for  background correction.           The user may want
     to have a look at different graphics (density plots, etc ...)  in
     order to decide what signal is more suitable to use. For details
     about signal processing see AFE User Guide.  'MeanSignal' is the
     spot Raw mean signal. 'ProcessedSignal' is the signal processed by
     the Agilent Feature Extraction image analysis software (AFE). It
     contains the  Multiplicatively Detrend Bacground Substracted
     Signal if the detrending is selected and it helps. If the
     detrending does not help, the 'ProcessedSignal' will be the
     Bacground Subtracted Signal.  'BGMedianSignal' is the Median local
     background signal.  'BGUsed' depends on the AFE software settings
     for the type of background method calculation and the setting for
     the spatial detrend. Usually, the Background Signal Used is the
     sum  of the local bacground + the spatial detrending surface value
     computed by  the AFE software. To view the values used to
     calculate this variable using different bakground signals and
     settings of spatial detrend and global background adjust, see
     Table 33 on page 213 of the AFE User Guide.  Limma function
     'backgroundCorrect' is used for the BG correction. This function
     is designed to produce positive intensities. Any intensity value
     lower less than 0.5 is reset to be equal to 0.5. Additionally, a
     constant of 50 (normally) is used as a offset that it is added to
     the intensity values before the log transformation.  The propouse
     of this calculation is to shrunk the log ratios to zero at the
     lower intensities and thus to reduce the variability of log-ratios
     for low intensity spots. The optimal choice for the offset is the
     one which makes the variability of the log-ratios as constant as
     possible accross the range of intensity values (Smyth, G. in BioC
     mailing List).   If the 'half' method is chosen for Background
     Correction, the method will substract the chosen BACKGROUND signal
     to the chosen FOREGROUND signal, to produce positive corrected
     intensities according to the 'half' method. If the 'normexp'
     method  is selected, then a convolution of normal and exponential
     distributions is fitted to foreground intensities using 
     background intensities as a covariate, and the expected signal
     given the observed foreground becomes the corrected intensity. See
     'limma' user guide for details.

_V_a_l_u_e:

     a 'RGList' object, containing in 'RGList$G' the log-2 normalized
     intensities

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

     Pedro Lopez-Romero

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

     Bolstad, B. M. (2001), Probe level quantile normalization of high
     density oligonucleotide array data. Unpublished Manuscript: <URL:
     http://bmbolstad.com/stuff/qnorm.pdf       >

     Bolstad, B. M., Irizarry R. A., Astrand, M., and Speed, T. P.
     (2003), A comparison of normalization methods for high density
     oligonucleotide array data based on bias and variance.
     Bioinformatics 19, 185-193.

     Smyth, G. K. (2005). Limma: linear models for microarray data. In:
      'Bioinformatics and Computational Biology Solutions Using R and 
     Bioconductor'. R. Gentleman, V. Carey, S. Dudoit, R. Irizarry,  W.
     Huber (eds), Springer, New York, pages 397 - 420

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

     Agilent Feature Extraction Reference Guide <URL:
     http://www.Agilent.com> See also 'backgroundCorrect' and
     'normalizeBetweenArrays' in the limma package and 'vsn' in the vsn
     package.

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

     ## Not run: 
              data(dd) 
              ddNORM=BGandNorm(dd,BGmethod='half',NORMmethod='quantile',
                             foreground='MeanSignal',background='BGMedianSignal',
                             offset=50,makePLOTpre=TRUE,makePLOTpost=TRUE)
     ## End(Not run)

