Gene (mRNA) - Protein Interaction databases and prediction

Question

[']

I have a set of proteins and I want to find out how they interact with a particular gene. Are there any tools and/or databases that would be suitable for this purpose?

Answer 1

[c.chakraborty]

I found these two- http://pri.hgc.jp/ http://lulab.life.tsinghua.edu.cn/postar2/rbp2.php But, I am not sure about reliability. Usually RBPs interact mRNAs with specific domains like RRMs, zinc-fingers etc. If I know which domain my protein has, I would data-mine literature to get probable RNA binding sequences. But if you have disordered regions in the protein or prion-like domains, it would be very hard to predict. You might look into these databases and tell us your opinion if it was useful or not! Also check the folding of the transcripts of this gene that you are testing using RegRNA http://regrna.mbc.nctu.edu.tw/html/prediction.html or mfold http://unafold.rna.albany.edu/?q=mfold. These tools are predictive, but hey if you're investigations turns out useful, you can use it to set an experiment to answer the protein-RNA interaction biologically.

[']

That's exactly the aim: predict computationally (for example RNA-binding capabilities between those proteins and the gene of interest), and if the result is reliable, start an experiment. Are there any known approach to do the prediction?

[c.chakraborty]

I am not really sure if there is one golden standard rule to do it. But, if I have to do this, I will find what motifs do my RBPs have and search literature for the sequences that these motifs can bind to, say a motif recognizes AU rich sequences. Then I record the sequence of that motif or motifs in r in a vector, and then search the transcripts for each motif RBP motif I have discovered, by using matchPattern from Biostrings. This would be a very preliminary and dirty way but would give me a glimpse of which proteins might be interesting. Also this will work if and if the RBPs have RRMs or AUbinding element, or polyA binding motifs. I mean RBPs known to bind sequences in the 3'UTR or ORF of the RNA, but if the RBP binds a structure, it will be very hard to predict. You can use mfold to see how your RNA folds (which might not be true in real) and then the PRD to check if your protein recognises the fold. But the Protein-RNA recognition is so complicated, it might be very difficult. But try and you may be lucky and successful.