8.9 years ago by
Maastricht, The Netherlands
This question relates to a lot of things indeed.
Macromolecular interactions are studied a lot and our understanding of how gene products work together is the basis of what we know about biological functionality as it is for instance present in regulatory or signaling pathways. And when it is about using this best known well structured knowledge about macromolecular interactions you will probably want to use pathway tools. There are a lot of questions on Biostar about these.
But you are right that the experimental finding of a protein-protein or a protein-nucleotide sequence by itself offers interesting information. Many databases exist for that type of information, for instance [?]IntAct[?] for protein protein interactions and [?]Jaspar[?] for protein nucleotide motif interactions. But there are really many many more, including some specific for protein RNA interactions. You will often want experimentally validated interactions but prediction is useful to know what needs to be validated.
You can try to combine interaction data from different sources. You will often want to start by taking a well known core, an existing pathway for instance, and extend from that using other data sources about validated or predicted interactions and to combine that with experimental data (for instance about co-expression or epigenetic regulation or just locations where proteins are found.). A good tool to do that is [?]Cytoscape[?]. And if you want to do network analysis often you will definitely want to learn how to use that.
Your specific structural biology background is of course very important to improve the predictions and you will want to evaluate the increased power of your predictions using approaches like I mentioned in the previous paragraph.
In fact these predictions are going to be a lot more important in an era where genetic variations can be easily measured. Because using structural biology you cannot only predict macromolecular interactions you can also evaluate how genetic changes would affect the macromolecules and thus the interactions. That really will be the core field where bioinformatics (or if you want integrative systems biology) has to develop over the next 10 years or so. Estimate how genetic variations affect transcription factor binding motifs, or protein binding domains... The questions go on and on. And a lot of it will need to be experimentally validated.
You might also want to check out [?]this page[?] with a recent webcast and a interesting overview on a Nature poster provided by Agilent.
I hope this helps a bit. Because you could easily write a book about it.
modified 8.9 years ago
8.9 years ago by
Chris Evelo ♦ 10.0k