[BioC] error in dispersion estimate

[mariangela bonizzoni]

Hi 
I am trying to use DESeq to test for differential expression between two 
conditions "DE" and "un". I do NOT have biological replicates. I have 7 
comparisons.

 conds <- c("CTM.14.car.un", "CTM.14.car.DE", "CTM.14.mid.un" , "CTM.14.mid.DE", 
"CTM.14.sal.un", "CTM.14.sal.DE", "CTM.01.car.un" , "CTM.01.car.DE", 
"CTM.01.mid.un", "CTM.01.mid.DE",
 "CTM.04.car.un" , "CTM.04.car.DE", "CTM.04.mid.un", "CTM.04.mid.DE")

When i set:


cds.CTM.01.car <- cds[ , c("CTM.01.car.un", "CTM.01.car.DE") ]
 cds.CTM.01.car <- estimateSizeFactors( cds.CTM.01.car )
 cds.CTM.01.car <- estimateDispersions( cds.CTM.01.car, method="blind", 
sharingMode="fit-only" )
 res.CTM.01.car <- nbinomTest( cds.CTM.01.car, "CTM.01.car.un", "CTM.01.car.DE" 
)


it works fine, (the same for CTM.14.car, CTM.01.mid)

but when i set:

cds.CTM.04.car <- cds[ , c("CTM.04.car.un", "CTM.04.car.DE") ]
 cds.CTM.04.car <- estimateSizeFactors( cds.CTM.04.car )
 cds.CTM.04.car <- estimateDispersions( cds.CTM.04.car, method="blind", 
sharingMode="fit-only" )


i get the error message (the same for CTM.04.car, CTM.04.mid, CTM.14.mid, 
CTM.14.sal):

Error in parametricDispersionFit(means, disps) : 
  Parametric dispersion fit failed. Try a local fit and/or a pooled estimation. 
(See '?estimateDispersions')
In addition: Warning message:
glm.fit: algorithm did not converge 


The option pooled was not liked because i do not have replicates. Any solutions? 
thanks