Cufflinks performs poorly with biological variance

[Apis37]

Hi,

I may just not have found the correct set of options, but Cufflinks performs very poorly with my data to the extent that I don't think the program is useful. Here is an example of a gene that is obviously differentially expressed (and is found to be so by other programs such as DESeq and by real time PCR) but which is not by Cufflinks.

This is the test:
value_1 value_2 log2(fold_change) test_stat p_value q_value
439.199 14731.1 5.06784 -1.01836 0.308507 0.999918

Here's the data calculated by cufflinks:
R1_count 2188.98
R1_count_variance 549561
R1_count_uncertainty_var 9.65605
R1_count_dispersion_var 319430
R1_status OK
R2_count 76824.8
R2_count_variance 1.09055e+11
R2_count_uncertainty_var 248.962
R2_count_dispersion_var 8.00216e+09
R2_status OK

Here is the output from HTSeq with the same data (raw counts):
R1 1: 546
R1 2 : 1993
R2 1: 35510
R2 2: 40164

There are many cases like this in the data I've tried to analyze with Cufflinks. I've tried several combinations of options and always get the same thing. Any advice would be greatly appreciated, otherwise I'll just give up on the program.

thanks