lncRNA expression association to gene

Question

[Pin.Bioinf]

Hello I have little knowledge about lncRNAs, and a collague asked if I can help him associate his lncRNA candidate to a gene of interest (a suposed target) by using rnaseq data he has. So he has provided me with the expression of the transcripts for different samples.

Would checking on the expression variation across samples for each, the lncRNA and gene(target) be enough? Even if their expression pattern was the same, I believe this is not enough to associate them, or is it?

Any other suggestions on what I could do? What do people usually do to associate a lncRNA to a target?

Thank you very much.

Answer 1

[Kevin Blighe]

The only way to definitively prove that a long non-coding RNA is interacting with a gene (i.e. the DNA) )or gene product (mRNA or protein) is through wet lab experimentation, of course. No in silico method can faithfully determine this because the number of variables that are required to be known for this are mostly unknown and un-measurable at present.

What people have done is to compute pairwise correlation matrices across all genes and then pick the correlations that are highly statistically significant. Doing this on a single CPU / thread for all of your genes (depending on how many genes you have...) is unfeasible, but I developed some parallelised code that can do it: https://github.com/kevinblighe/clusGapKB/blob/master/corKB.R

That code just generates the Parson correlation value. To develop the P value, use the cor.test() function.

----------------------------------------

For other tools, a quick search reveals a question posted on ResearchGate: https://www.researchgate.net/post/Any_recommendation_of_lncRNA_target_prediction

You will also be interested in reading this:

Bioinformatics tools for lncRNA research

Kevin