Dear TMEV development team,
I am running TMEV 4.6.1 build 2443, downloaded and installed as recommended on laptop with WinXP/SP3.
Input file is .tdms/txt, with columns gene ID, condition 1 (4 replicates), condition 2 (4 replicates), condition 3 (4 replicates), condition 4 (3 replicates), and rows are 9553 probes with condition/reference ratio, lowess-normalized and log2-transformed, present in all conditions and all replicates.
SAM input parameters are always the same:
S0=Tusher et al method; Calculate q-values=Yes; Use calculated number of permutations.
When I run SAM in non-“Use R” mode for one class and two class unpaired analysis, in output table views all columns have corresponding meaningful values.
When I run SAM in “Use R” mode for one class and two class unpaired analysis:
1. In output->table views output only zeroes are written in columns Numerator(r), Denominator(s=s0), and q-value (%), although Expected score(dExp) and Observed score(d) are calculated;
2. In output->General information->Computed Quantities: Computed Exchangeability Factor s0: 0.0 (which is unlikely).
Are there problems with reading from samr output tables to TMEV output table views?
Also, when I analyse the same data and use exactly the same cut-off parameter which is median number of false genes =0. SAM and samr output are different in calculated delta and its upper and lower cut-off values, and, consequently, number of differentially expressed genes.
Are there known differences between SAM-1.0 and samr-1.27 implementation of calculation algorithms?
P.S. I have similar problem with LIMMA outputs, that I reported in thread "LIMMA questions" - are there anybody looking at it?