I am wondering whether it is completely necessary to have a biological replicates for an RNA seq experiment to be prepared on the same day. What if in case of a differential gene expression analysis, one replicate in both control and case was prepared on different day from other samples. Will it make the data analysis hard and impossible?
Hi Srimeenakshi,
preparing samples in different days (and therefore different conditions) could led you to biased data due to group effects. or maybe you won't have choice due to the amount of samples.
fred
It depends on what do you mean by processing. If it is RNA extraction and library prperation then it is alright. If you meant growing cells, treating, harvesting cells/tissues etc. then you will see a huge variability of course as other mentioned. However, I believe that is biological replication in a true sense. If someone cant repeat what you saw under same conditions then it is not what you describe it is. As long as you are careful not to vary growth conditions, processing times/ methods (if at all possible), then the noise you see will be biological noise under your experimental setup. Of course it will be tough to capture many diiferential expressed genes but you might capture true drivers of change under these conditions. That is up to you to decide depends on how noisy your experimental system and biological process you are looking at etc. If it is, then may be a good idea to process then all together.
Thank you very much !!!!
Frederic Lepretre Hua Xiao Ajit kumar Roy
Shajahan Anver
I assume you are asking about the sample preparation of library preparation.
It depends on the design of the experiment. It would be the best to do all the samples from all the conditions in one day, with the same 'batch of chemicals', same pipette, same temperature, same time, etc. If this is not possible I would randomize the samples as much as possible - prepare some biological replicates from both of the conditions on one day and some on the other day. It is very likely you will get some technical batch effect during the sample preparation which is very simple to remove during the analysis only in case it's the same for all the conditions. If it would be specific only for one condition it is impossible to evaluate.
Let me give you an example: you prepare healthy samples on Monday and cancer samples on Tuesday. You use a different 'batch of chemicals' each day. The different batch of chemicals and preparation day may (or may not) cause a technical batch effect. During the differential expression, you see a difference which you cannot be sure is caused by the different conditions or simply by a different batch of chemicals/preparation day. You also cannot remove it from the experiment (during differential expression calculation) as you cannot know whether you are removing a real batch effect or the biological differences between the conditions. If you prepare half of the samples from all the conditions on one day and half the other day (half healthy and half cancer on Monday and half healthy and half cancer on Tuesday) you can include the batch of chemicals/day information in the differential expression formula and since the batch effect would be the same for both of the conditions you can remove it quite easily.
To sum it up - the more you randomize the more you can be sure the differences you are observing are actually caused by the biological differences and not by the technical differences. It also applies to the sequencing itself. If possible, do all the samples the same, with same chemicals, on the same day. Of course, there are some exceptions, as always.
- Badges
- Science topic