-nan for FPKM value in Cuffdiff output

Question

[mccormack]

Some recent gene_exp.diff files, the output of Cuffdiff, have '-nan' for some Value 2 FPKM values. I do not see how this can happen because I would think you could only have only a '0' or a positive value for FPKM. This only seems to happen for value 2 and not in the value 1 FPKM column. Has anyone seen -nan in the FPKM column?

Answer 1

[DG]

As in the columns value1, value2? No I don't think I have typically seen this in my data. Usually you get NaN or -nan, etc where not expected due to overflow or underflow errors. I typically only see it in my Diff files inplace of a P-value when the FPKM of one of the samples is 0 for that particular transcript, but that is expected behaviour.