C5h C10h C15h T5h T10h T15h
gene1 1 2 3 30 32 29
gene2 10 12 13 20 21 23
gene3 100 103 105 200 205 208
gene4 1000 1100 1200 1600 1600 1650
I got the RPKM value of members in a gene family (encoding an enzyme), as showed above, how to predict the importance of these members(C:control ; T:treatment; h:hours)?
I have tried PCA (principal component analysis )to find members with representative expression pattern as well as high absolute correlation coefficient with PC1 which was representative of the major variance, but often low RPKM.
so I was confused, is there any method that can select the members with high RPKM as well as difference between control and treatment groups?
any questions and suggestions will be very appreciated.
Hi,
Why don't you perform a differential expression analysis to determine which changes (treatment vs control) are significant and which not?
You can also consider the time component to define the group of genes with significant change and the group with normal variation.
Best,
Juan
Ok, so we start from the basis that all those genes present differential expression.
Apart from PCA you could use traditional clustering algorithms to find groups of genes. Now, if you want to get groups also based on diff. control vs treatment, I would add additional columns with these values (fold change). In this case then consider scaling the values.
I hope that helps.
Best,
- Badges
- Science topic
- Similar topics
- Comment