I have several bacterial strains that I am generating RNAseq libraries for. I have collected 4 biological replicates of each, in sets where all of the strains were grown on the same day. All of the bio reps were grown and collected with as little variation as possible, including using the same bottle of media that has now been used up.
Of course I screwed up one of the replicates. The cell pellets sat on the bench for too long before going into the bead beater to be lysed, and my total RNA yield dropped by 30-70% (from 10 ug usually obtained).
So now I have the dilemma - I still have enough RNA from each sample in the damaged set that I could technically create a library, but I am not sure if it is worth the cost of the reagents necessary to do so or if the data would be reliable.
I could technically generate a new replicate set, but it will probably have a lot of extra variation compared to the other bio reps because it will have to be grown with a new bottle of media.
I could also just go ahead with 3 biological replicates instead of 4, but I am reluctant to lose the statistical power.
Does anyone here who has examined their RNAseq data have any opinion they'd be willing to share as to which option may better? I know there is no optimal option, but I would like to salvage the situation in the best manner.
Of course I screwed up one of the replicates. The cell pellets sat on the bench for too long before going into the bead beater to be lysed, and my total RNA yield dropped by 30-70% (from 10 ug usually obtained).
So now I have the dilemma - I still have enough RNA from each sample in the damaged set that I could technically create a library, but I am not sure if it is worth the cost of the reagents necessary to do so or if the data would be reliable.
I could technically generate a new replicate set, but it will probably have a lot of extra variation compared to the other bio reps because it will have to be grown with a new bottle of media.
I could also just go ahead with 3 biological replicates instead of 4, but I am reluctant to lose the statistical power.
Does anyone here who has examined their RNAseq data have any opinion they'd be willing to share as to which option may better? I know there is no optimal option, but I would like to salvage the situation in the best manner.