Bioinformatics (Applications in Pharmacology)

[InSilico]

I've recently developed a fascination in bioinformatics. The idea of using in silico models for things like hit and lead identification, screening millions of compounds for similarity in shape or pharmacophoric activity to discover a handful that might have potential for further tests (in vitro and in vivo) - it's something that I see as having an enormous potential. Of course, the applications of bioinformatic tools don't end with lead identification. Theoretically, one might be able to use digital simulations to test for alternative interactions between the lead compound (drug candidate) and other, non-intended (non-target) endogenous sites (proteins, DNAs, RNAs, for example). This, then, might be useful in lead optimization, where you'd then have to manipulate the compound in order to increase specificity and decrease toxicity.

 

So you understand that I think this is a cool set of concepts. The notions here excite me, but I'm really interested in learning more. As much more as I possibly can. I realize that the things I've described above are conjecture based on ideal databases and ideal tools which may either not (yet) exist or may not be available to anyone but their developers. I'm wondering, though, quite how deep this rabbit hole goes and what I can learn. If you have any background in the field, or maybe if you know somebody who does, you'd be doing me a tremendous service in helping me expand my understanding of the role of bioinformatics in the drug discovery and drug development process.

 

I've been told that these notions have existed for decades; that the concept of in silico screening in drug discovery has existed for a long time now and hasn't changed much. I don't know what to make of that claim. Is it true? If no significant advances have been made in the last, say, 5 years, why not? With the advances in computing power, I would imagine that the field of bioinformatics would be evolving in direct relation. Also, can anyone identify an instance of successful drug design (that is, a pharmaceutical that was demonstrated as effective and/or marketable) that relied on bioinformatic screening techniques? Is there a drug out there that was designed from such a rational approach? I'd love to get some real-world examples I can explore further.

 

I want to say finally that I appreciate a community like this and all of you who enable it. I've been a member here in the past (under a different email and username which I'm afraid I've since forgotten), and I regret having been away for so long. I hope to generate meaningful discussions now that I'm back. I'm looking forward to getting to know you all, again.

 

Thanks in advance for your input.

[CharonY]

Not my field, the little I have seen were PhD projects in the area of in silico ligand-recptor docking (also with pharma application in mind). The impression I got from what I have seen implies that the predictions work decent for well-resolved and understood structures (i.e. where plenty of information is available). However, structural information appears to be at least one of the limiting factors. It is more used for drug modification rather than de novo drug discovery for this and other reasons. Toxicity effects are even more complicated as it is not clear where other reaction may occur and then the complexity increases exponentially. Then problems as lack of data, lack of structural information, complexity of interaction and overall lack of integrative models become dominant.

 

 

So in short, from the little I have seen it works in small, but little progress has been made in order to understand more complex interactions (which, incidentally, is closer to my area of expertise).

Edited September 22, 2011 by CharonY

[InSilico]

The notion that there haven't been big advances in the understanding of complex molecular interaction is a excuse for a lack of progress in the computational drug design world, and it's something I figured was the case, but wasn't able to find any real citations to uphold it. Can anyone else confirm (conclusively) that this is really the cause for the holdup?

 

So in short, from the little I have seen it works in small, but little progress has been made in order to understand more complex interactions (which, incidentally, is closer to my area of expertise).

 

Oh? Sounds neat. What all do you study?

[CharonY]

The notion that there haven't been big advances in the understanding of complex molecular interaction is a excuse for a lack of progress in the computational drug design world, and it's something I figured was the case, but wasn't able to find any real citations to uphold it. Can anyone else confirm (conclusively) that this is really the cause for the holdup?

 

 

Well, I am pretty sure that it is being mentioned in several reviews. However, this is more of a common knowledge thing, considering the large efforts that are still required to obtain structural information (even inaccurate ones). The mantra of computational modelers in this area tend to be "gimme a target, gimme target". Note that the problems are mostly on the biological side (e.g. receptors). The reason is their intrinsic complexity that it takes huge efforts and time to gain structural information from proteins. Now, considering that the analysis of an 1:1 interaction is already tricky, you can imagine how big of a pain it is to model more interaction (and how impossible it gets if you consider how few structures in the end we got).

Even worse, understanding the outcome of interactions requires even more biological knowledge that, to a big part, is still lacking, which would help in prioritizing targets.

This again is, in my opinion, related to the fact that most funding goes into the direct drug target identification, but less is invested in trying to understand fundamental cellular processes, but I digress.