home | about | Oxford University | research | groups | public | links & resources | news | seminars | vacancies | contact | find us

This page describes the output of the C version of HAPPY. Refer to the PDF documentation for the R version.

HAPPY analyses each marker interval separately, fitting a linear additive model for trait effects assuming a QTL is present within the interval. An effect size is estimated for each ancestral strain, and the hypothesis that there are no significant differences between the strain effects (ie no QTL in the interval) tested by ANOVA F statistic. A typical output for one interval looks like this:

     Testing marker interval 11,12  D1MIT264 D1MIT194 
     strain densities:
             A/J        AKR       BALB        C3H        C57        DBA          I       RIII
          0.1891     0.5380     0.2572     0.1896     0.2575     0.0778     0.1691     0.3217
     ANOVA F 5.963372e+00 pval 9.087284e-07
     tss 4.520509e+02 fss 2.404637e+01 rss 4.280046e+02 R^2 5.319394e-02 
     trait estimates, mean= 1.469828e-01:
      1        A/J effect -8.958640e+00 se  4.572822e+00 T -1.959105e+00
      2        AKR effect -2.302646e-02 se  8.002267e-02 T -2.877492e-01
      3       BALB effect -7.967934e+00 se  1.104530e+01 T -7.213870e-01
      4        C3H effect  9.463201e+00 se  4.479047e+00 T  2.112771e+00
      5        C57 effect  7.735892e+00 se  1.102880e+01 T  7.014264e-01
      6        DBA effect  9.941300e-01 se  3.721886e-01 T  2.671038e+00
      7          I effect -3.325929e-01 se  5.296803e-01 T -6.279125e-01
      8       RIII effect -6.170642e-01 se  1.919972e-01 T -3.213924e+00
     

• This means the marker interval between markers D1MIT264 and D1MIT194 is being tested.
• The strain densities tell you what proportion of ancestral chromosomes are predicted to be from each strain over the interval (thses numbers always sum to 2 (as the individuals are diploid).
• ANOVA F is the F statistic for the analysis of variance, pval the corresponding p-value. tss is the total sum of squares, fss the fitting sum of sqaures, rss the residual sum of squares, R^2 the percentage of total variance explained.
• The trait estimates are given for each strain, eg

   1        A/J effect -8.958640e+00 se  4.572822e+00 T -1.959105e+00
  

  means strain A/J has effect size -8.95, standard error 4.57, T-values -1.95. Not that because sometimes strain effects are highly correlated (in regions where two strains cannot be distinguished), it is not advisable to interpret the significance of individual strain effects using a T-test; it is safer to look at the F-statistic.