Hi
I have 3 High-throughput RNA-Seq samples from patients with rare disease and 3 samples from healthy subjects. I have identified the differentially expressed lncRNA and coding-genes between patients and control subjects. I would like to predict the function of identified differentially expressed lncRNA based on the expression of coding genes. What is the best method for this goal ?
I found that one way is to calculate the Pearson's correlation coefficient (PCC) between the expression values of each of the lncRNA-mRNA pairs.
I would like to know is this approach valid for my sample size ( 3 patients) ?
Should I just calculate the Pearson correlation confidence base between the expression values of each of the lncRNA-mRNA pairs only across patients samples?
Thanks
•
written
by
hana
Are those LncRNAs are novel and tissue specific ? Its not valid to do a correlation with 3 samples.
If the LncRNAs are not highly tissue specific, you can gather RNA-Seq data from all resources (GEO, GTEx, bodymap etc where the LncRNAs show a minimum of 1 FPKM/RPM expression) and do a pairwise co-expression correlations to find which genes are most correlate with the expression of LncRNA. This will help to prioritize your differentially expressed genes that are regulated by differentially regulated LncRNAs.
P.S: I just realized its very old post :-/ . Still better to leave an answer.
P.S: I just realized its very old post :-/ . Still better to leave an answer.
Caught me out too ;) Upvoted to help others and prevent this question from being 'bumped' again by the Biostars bot.
Log in to answer this question.
you can do like this
yan i think you forgot link
Would you please send the link
Hi Hana,could you help me?our works are somehow similiar
3 samples is just not really enough for the purposes of constructing a network. You would have a difficult time convincing anyone of the results / inferences from the network.
Could you please you share the link you wanted to post ? Or kind of approach have you undertaken? Thanks