I can only agree with h.mon.
I recently conducted a study involving a bacterium. I de novo assembled the transcriptome and was able to identify key transcripts that were differentially expressed between the two strains of the bacterium. There are tools like blastx that can help you to understand the functionality of 'unknown' sequences.
To de novo assemble the transcriptome, there are various options:
Velvet followed by Oases (https://www.ebi.ac.uk/~zerbino/velvet/)
TopHat2 / Cufflinks to produce a FASTA reference, and then Kallisto to count abundnces
Trinity (https://github.com/trinityrnaseq/trinityrnaseq/wiki)
Rockhopper (https://cs.wellesley.edu/~btjaden/Rockhopper/)