Appropriate number of reads for assembling novel transcripts with Trinity?

Question

[alex.rubinsteyn]

I'd like to detect novel transcripts in a cancer sample with Trinity. I'm sequencing on an Illumina Hi-Seq (RiboZero capture, 150bp paired end reads). Are there any rules of thumb for how many reads I'll need to sequence to confidently assembly a majority of the transcripts?

[dario.garvan]

How would you know if you found the majority of novel transcripts if you don't know how many there are ? If you only found one novel transcript which was the cause of the cancer you are studying, isn't that also desirable ?