samrocN                 package:SAGx                 R Documentation

_C_a_l_c_u_l_a_t_e _R_O_C _c_u_r_v_e _b_a_s_e_d _S_A_M _s_t_a_t_i_s_t_i_c

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

     Calculation of the regularised t-statistic which minimises  the
     false positive and false negative rates.

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

     samrocN(data=M,formula=~as.factor(g), contrast=c(0,1), N = c(50, 100, 200, 300),B=100, perc = 0.6, 
      smooth = FALSE, w = 1, measure = "euclid", p0 = NULL, probeset = NULL)

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

    data: The data matrix, or ExpressionSet

 formula: a linear model formula

contrast: the contrast to be estimnated 

       N: the size of top lists under consideration

       B: the number of bootstrap iterations

    perc: the largest eligible percentile of SE to be used as fudge
          factor

  smooth: if TRUE, the std will be estimated as a smooth function of
          expression level

       w: the relative weight of false positives

 measure: the goodness criterion

      p0: the proportion unchanged probesets; if NULL p0 will be
          estimated

probeset: probeset ids;if NULL then "probeset 1", "probeset 2", ... are
          used.

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

     The test statistic is based on the one in Tusher et al (2001):


                         d = diff / (s_0 + s)


     where diff is a the estimate of a constrast, s_0 is the
     regularizing constant  and s the standard error.  At the heart of
     the method lies an estimate of the false negative and false
     positive rates. The test is calibrated so that these are
     minimised. For calculation of p-values a bootstrap procedure is
     invoked. Further details are given in Broberg (2003). Note that
     the definition of p-values follows that in Davison and Hinkley
     (1997), in order to avoid p-values that equal zero.

     The p-values are calculated through permuting the residuals
     obtained from the null model, assuming that this corresponds to
     the full model  except for the parameter being tested,
     coresponding to the contrast  coefficient not equal to zero. This
     means that factors not tested are kept fixed. NB This may be
     adequate for testing a factor with two levels or a regression
     coefficient (correlation), but it is not adequate for all linear
     models.

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

     An object of class samroc.result.

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

     Per Broberg

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

     Tusher, V.G., Tibshirani, R., and Chu, G. (2001) Significance
     analysis of microarrays applied to the ionizing radiation
     response. _PNAS_ Vol. 98, no.9, pp. 5116-5121

     Broberg, P. (2002) Ranking genes with respect to differential
     expression , <URL:
     http://genomebiology.com/2002/3/9/preprint/0007>

     Broberg. P: Statistical methods for ranking differentially
     expressed genes. Genome Biology 2003, 4:R41 <URL: 
     http://genomebiology.com/2003/4/6/R41>

     Davison A.C. and Hinkley D.V. (1997) Bootstrap Methods and Their
     Application. Cambridge University Press

