If I have a single replicate per condition and I am trying to see if the differences in gene expression ( for a single gene) are significant between conditions what would be the correct test to use? Also out of curiosity, if some conditions have a single replicate while others have more than one, which test should be used then?
Thank you
•
written
by
Tonkatsu •
1 answer
Without replicates, you don't get any idea about the underlying distribution of values so you can only say if two numbers are equal or if one is greater or smaller than another. If you have replicates, you can get an idea of the underlying distribution for that condition and then you can try to build a prediction interval but unless you can assume some parametric distribution (e.g. the data is normally distributed), in the non-parametric case, you'll probably need quite a few replicates.
Log in to answer this question.
-
Statistical test of difference between two conditions and account for the batch
written by npont •
-
Am I accounting for replicates correctly in pseudobulk analysis?
written by Aspire
-
Using Pseudobulk Approach for Identifying Marker Genes Within a Single Condition
written by nhaus •
-
qPCR gene expression data analysis
written by ebertomeup •
-
GSEA vs Functional Annotation
written by turcoa1 •
-
STAR alignment and HTseq-count
written by otieno43 •
-
DESeq2 comparisons using contrast
written by basuanubhav •
-
Normalization for comparing expression statistics across sample and gene groups
written by endre.sebestyen •
-
Forum: How to convincingly illustrate and discuss negative results in NGS.
written by Carlo Yague
-
statistical test on count data for differential expression
written by tonja.r
You're talking about comparing two numbers in a statistically significant way. I don't think that's possible.
For the second question, you may be looking for a t-test. Read more here https://compgenomr.github.io/book/how-to-test-for-differences-between-samples.html