Besides the second reason given by Swaraj (the first is not relevant if you have sequenced the genome), there are three main reasons to use paired-end sequencing of RNA for prokaryotes.

First, it gives you a better "coverage" of your transcripts. In contrast to eukaryotes with splicing, you can be fairly sure that the piece in between the mapped read pair is present in your cDNA, so you can create a fake read covering the whole piece of cDNA.

Second, you get a better resolution of the 3'-ed of your transcript. That can help you to better define 3'-UTRs and novel ncRNAs where you cannot infer the likely end from a stop codon.

Third, it allows you to better define polycistronic mRNAs and thus operons.

Last but not least: Back when sequencing length was still an issue, you could get all the above from sequencing just 2x 20-25bp (usually more than enigh to properly map a read within a bacterial chromosome).

Oh, and if you do metatranscriptomics, paired end data will help you tremendously to assemble unknown transcripts.

The two most important aspects I can think of, considering prokaryotic genomes are:

- Paired end gives an idea of the size of the insert and the diectionality of the mapping to the sequence assembly algorithms. This aids in prediction of inversions, deletions and mutations inside the genome.

- High quality of mapping of reads in regions with repeat content, Single end reads are woefully unsuitable to accurately predict transcription in repeat containing regions.

Hope this helps.

Here's the complete explanation.

http://rnaseq.uoregon.edu/exp_design.html

Besides the second reason given by Swaraj (the first is not relevant if you have sequenced the genome), there are three main reasons to use paired-end sequencing of RNA for prokaryotes.

First, it gives you a better "coverage" of your transcripts. In contrast to eukaryotes with splicing, you can be fairly sure that the piece in between the mapped read pair is present in your cDNA, so you can create a fake read covering the whole piece of cDNA.

Second, you get a better resolution of the 3'-ed of your transcript. That can help you to better define 3'-UTRs and novel ncRNAs where you cannot infer the likely end from a stop codon.

Third, it allows you to better define polycistronic mRNAs and thus operons.

Last but not least: Back when sequencing length was still an issue, you could get all the above from sequencing just 2x 20-25bp (usually more than enigh to properly map a read within a bacterial chromosome).

Oh, and if you do metatranscriptomics, paired end data will help you tremendously to assemble unknown transcripts.

When mapping single-end reads you could have some problems for genes that are located in the same genomic region but in different strands, since you don't know from which strand the reads belong. I think that paired-end reads could be helpful in these issues, especially for prokaryotes where the transcripts are densely distributed in the genome.

Swaraj wrote:

> Paired end gives an idea of [ ... ] the diectionality of the mapping to the sequence assembly

This is only true if you use a strand-specific protocol (which is strongly advised for RNAseq anyway), and in that case, both SE and PE will tell you the directionality.