problems with zero counts using cufflinks

[yaaraore]

I encountered a problem during my last run of cufflinks. From some resons I got that when sample1 is zero and the other is a number (even a very high one like 7000) it returns that the test was not significant. I ran a different genome with the same parameters and this time I did not get this problem (I ran versions 1.3.0 and 1.2.1).
Any idea why something like this happens?

[jmwhitha]

Did you ever figure this out? I am also having this problem. My p-value and q-values are defaulting to 1 while my test_stat is defaulting to zero for value_1 or value_2 = 0 or close to zero compared to very high values.

In addition to this I am getting LOWDATA status despite clearly very high expression (160k count vs 850 count, for example).

Any help would be much appreciated!
Jason