Combining >two mappings (reference and de novo) in hexaploid wheat

[Gordona]

Hi there,
Has anyone combined more than one mapping data set when analysing with DESeq2 or EdgeR?

We are looking at the interaction of a fungus and hexaploid wheat as an infection takes place and have sequenced using Illumina 100bp SE reads.

We have the genome sequenced of the fungus and have used the predicted coding sequences as a "reference genome" to map to, and have used the NCBI Unigene set (of 168 000 genes) for the wheat "reference genome".
We have just done a de novo assembly of the leftover unmapped reads and have a new set of contigs that we've used as a third "reference genome" and now have mapped count data for that.

Are there major issues with combining these datasets and running DE analysis in one go?

When the de novo contigs (729000) are blasted against the NCBI wheat unigene set, 321000 align, suggesting that we've generated contigs for the two other homeologues. Has anyone else used the Wheat Unigene set and found the same? How have you analysed homeologous gene expression?

Any advice welcome,
Many thanks,

Anna