mapping total RNA to a subset of the genome

[mrbig]

I'm interested in the differential expression of organellar genes (i.e. chloroplast and mitochondrial genes) in a particular plant species under salt treatment vs fresh water treatment in two tissue types (leaf and root). We generated four paired-end EST datasets from total RNA (i.e. RNA from all three plant genomes):

1. leaf-sample with salt water
2. leaf-sample with fresh water
3. root-sample with salt water
4. root-sample with fresh water


Unfortunately, no technical or biological replicates were made.

The organellar genomes of this plant species have been published. However, the nuclear genome is not yet available.

So I have mapped the datasets onto the organellar gene coding sequences. I then counted the number of reads that map to each gene. Here's what some of the results look like:

LEAF:
gene salt no-salt
psbA 30856 1882
matK 294 62
rbcL 12472 3547
atpB 525034 497024
atpE 259993 280664
ndhC 4172 2408
ndhK 11702 7912
ndhJ 9439 6931
.
.


ROOT:
gene salt no-salt
psbA 133 69
matK 17 4
rbcL 297 428
atpB 12755 16277
atpE 9011 11622
ndhC 102 175
ndhK 463 548
ndhJ 541 511
.
.


My question is, what sort of statistical test or normalization do I need to perform to show the effect of the salt given that (a) I have no biological replicates and (b) I have mapped total RNA reads (including nuclear reads!) against a subset of the genome (the chloroplast and mitochondria)?

Any help will be very much appreciated. Cheers.

[mrbig]

The question is ... what proportion of chloroplast to mitochondrial to nuclear reads is there within each dataset? And no doubt the proportion of reads will be different between each dataset.

Would a Fisher's exact test help? If so, how can I apply this?