Hey,

the question shows that you have very limited experience in this field, I hope I can help you out a little bit.

First of all, the question is NOT how many reads you can get but more "how many reads do you need?". Usually you will aim for 40-50 million reads per sample if you are sequencing human mRNA. BUT: you can always go for more until there is no more space on the flowcell or the given sequencing device. You should take a look at the Illumina website for that.

Second: You should ask yourself what is the purpose of the sequencing. If you want to call SNPs you should probably aim for more reads. Are you aiming for fusion transcripts, you should probably use longer reads.

Last but not least: You should be sure about what you WANT to do and after that read some publications on Next Generation Sequencing to get a better feeling for what you CAN do.

Hope this helps!

Regards,

André

Answering to your question, the number of reads you get out depends on what platform you are using and how many samples you wanted to pool. Most commonly, RNA-seq is done on Hi-Seq platforms which has 8 lanes and each lane generate around 200 million reads and in a total of 2 billion paired end reads per run. This is just information about how much data you can generate.

And now why are you doing RNA-seq analysis.

As mention in previous answer the purpose need to be known to go for the number you would need per sample. For detail, I can elaborate more. First you need to think about your reference genome and how well genome of your plant is studied. Suppose if you are doing very deep RNA sequencing and there is not good genome, it would be hard to map the reads. Of course you can use model organism to map, but it is not really going to give you much detail of the rare transcript variants or any novel transcript.

For just checking the gene expression profile, a low read output (5-25 million) is fine and recommended. If you want to dig deeper for deep transcriptome analysis, you would need a 2-3 fold reads. Furthermore, I personally don't think SNP calling on RNA-seq data is a good idea, if the data is not complemented with DNA/genome analysis.

A more practical thing to mention is number of reads is not always mean the number of mRNA reads or number of mapped reads. There is always a fraction from rRNA. Thus, it is very much dependent what kind of rRNA depletion was carried out and how efficient it was. Thus RNA-seq experiment begins with the sample prep and not just putting samples on sequencer.

Number of reads obtained is purely related to sequencing technology and library preparation. With plant you have to be very careful about library rRNA and non-PolyA transcript filtering - mitochondrial and chloroplast transcripts can swamp the nuclear chromosome mRNA output dependent on time of sample collection in the plant growth cycle and environmental conditions. With polyploid plants, and usually aligning against incomplete targets containing many assembly artifacts, I would be going for paired reads (reads of at least 100bp and insert sizes around 150-180bp so as to give a small overlap) with very strict aligner parameterisation - allowing for at most 3 or 4 mismatches - and aligning to the complete target, not individual chromosomes. If targeting a 4GBp genome then I would be considering a sequencing run of around 500 million PEs relying on coverage depth to gain confidence in the alignments. Which gene copy and haplotype did that read really originate from????