Reads in tophat missing in cuffdiff

[wotfanak47]

I'm using cuffdiff2 to examine the behavior of two paralogs in different maize haplotypes, and I've run into some output that isn't consistent withe previous data on these genes. qRT-PCR on these genes shows expression and Tophat alignments of single-end Illumina reads show expression as well. However, when I run cuffdiff2 on these files (2 biological replicates per haplotype), gene1 is assigned non-zero FPKM values and gene2 is assigned zero FPKM in all haplotypes. The quantification status for both genes is OK in all instances. I've had issues before with cuffdiff not recognizing annotated genes, creating novel transcripts where annotated models exist, etc., so I looked for a "novel" transcript at the same locus as gene2 but found nothing. I'm rather stymied at this point, and I'm curious if it sounds like I'm doing something wrong, others are having a similar problem, how to fix it, suggestions.