I have RT-PCR data of siRNA silencing in order to check the expression of other genes wrt to the silenced gene. When I normalized the data wrt internal control (U6, 18S RNA) and used the student t-test for the statistical analysis, the t-test value was found to be significant between the Si-control and Si-target. On the other hand, when I showed the variation in the Si-control values (i.e., without normalization) and plotted alongside the Si-target values, the t-test value was not found to be significant.
Is it acceptable?
Can anybody guide me on this?
Because RT-qPCR data are compositional by experimental design (and for several other reasons, but it is one of the most stringent ones), comparing the raw data directly is useless, if not misleading, expect in very special cases.
You must use either normalized data, or data analysis methods that take into account the compositional nature of the data.
Note however that normalization by internal controls makes strong assumptions about these controls, that are not always satisfied; if not satisfied, results may be anything after « normalization »…
See for instance Curis & al., 2019, Bioinformatics [PMID: 30010788] & Curis et al., 2019, Scientific Reports [PMID: 31700128] for a discussion.
- Badges
- Science topic
- Similar topics
- Instrumentation Engineering
- Instruments