Sleuth: What is the correct function for fitting my sleuth object

Question

[scottschu97]

I am looking at gene expression between four different tissue types: macula, macRPE, tissue, and retina with conditions of either AMD or No AMD

I have the following design table:

 sample tissue  condition
 1   AMD_cornea.17 cornea        AMD
 2   AMD_macRPE.18 macRPE        AMD
 3   AMD_macula.11 macula        AMD
 4   AMD_macula.12 macula        AMD
 5   AMD_macula.14 macula        AMD
 6   AMD_macula.17 macula        AMD
 7   AMD_macula.18 macula        AMD
 8   AMD_macula.19 macula        AMD
 9   AMD_retina.11 retina        AMD
 10  AMD_retina.12 retina        AMD
 11  AMD_retina.14 retina        AMD
 12  AMD_retina.17 retina        AMD
 13  AMD_retina.18 retina        AMD
 14  AMD_retina.19 retina        AMD
 15  ctrl_cornea.1 cornea no.disease
 16 ctrl_cornea.10 cornea no.disease
 17 ctrl_cornea.15 cornea no.disease
 18  ctrl_cornea.2 cornea no.disease
 19  ctrl_cornea.3 cornea no.disease
 20  ctrl_cornea.7 cornea no.disease
 21 ctrl_macRPE.13 macRPE no.disease
 22 ctrl_macRPE.15 macRPE no.disease
 23 ctrl_macRPE.16 macRPE no.disease
 24  ctrl_macRPE.7 macRPE no.disease
 25  ctrl_macRPE.9 macRPE no.disease
 26 ctrl_macula.10 macula no.disease
 27 ctrl_macula.13 macula no.disease
 28 ctrl_macula.15 macula no.disease
 29 ctrl_macula.16 macula no.disease
 30  ctrl_macula.4 macula no.disease
 31  ctrl_macula.6 macula no.disease
 32  ctrl_macula.7 macula no.disease
 33  ctrl_macula.8 macula no.disease
 34  ctrl_macula.9 macula no.disease
 35 ctrl_retina.10 retina no.disease
 36 ctrl_retina.13 retina no.disease
 37 ctrl_retina.15 retina no.disease
 38 ctrl_retina.16 retina no.disease
 39  ctrl_retina.4 retina no.disease
 40  ctrl_retina.6 retina no.disease
 41  ctrl_retina.7 retina no.disease
 42  ctrl_retina.8 retina no.disease
 43  ctrl_retina.9 retina no.disease

I would like the model to account for both tissue and condition When fitting my sleuth object, which is the correct function? Should I use:

 so <- sleuth_fit(so, ~tissue + condition,'full')
 so <- sleuth_fit(so, ~tissue, 'reduced')

Or

 so <- sleuth_fit(so, ~tissue + condition,'full')
 so <- sleuth_fit(so, ~1, 'reduced')

Any help is appreciated!

Answer 1

[Devon Ryan]

Your reduced model should be ~tissue if you want to test the effect of condition and ~condition if you want to test for the effect of tissue.