Are there commercial opportunities available for providing computational services for small companies without financial resources to establish an in house wet-lab?
The semiconductor industry has an established model of fabless chip design companies. Are we going to see a similar model in the drug or enzyme design industry any time soon? From searching on the Internet I do not get the impression that such a business model prevails today. What are the barriers we need to over come before such a model becomes successful the same way as the fabless chip design model has prevailed?
Commercial Opportunities for computational services in bioinformatics
Mon, 2005-04-25 13:11 — Qasim Sheikh
User login
Navigation
New forum topics
Recommended Reading
Who's online
There are currently 0 users and 12 guests online.
Syndicate
Powered By
Comments
Computational services
My company (www.rmcbiosciences.com) offers services in Computer Aided Drug Design, which is basically computational sciences. Many small- to mid-size biopharmaceutical companies prefer to outsource their molecular and computational modeling because it is not part of their core competency. This is, however, a fairly small niche market.
Is computation the problem?
I'm assuming you're suggesting startups providing computational services rather than wet services? [The first para is ambiguous]
I don't know that the design phase is yet that routine in the drug/enzyme industries - at least to the point of being largely a matter of in silico prediction. Afaik, leads are usually gotten by either extensive screening, or semi-random modification of existing compounds (for which there is already a booming industry ).
How far can we go with just computations
In the 2nd paragraph of your response you mention "extensive screening, or semi-random modification". Are these steps done purely computationally? Are computational results useful to a drug designer or do we have to show some wetlab results to support the validity of these computational results?
validation
Afaik, it's done in vitro. Rational drug design is still very much in its infancy, and there is simply no way to predict the effects of a certain strucuture - at least not well enough to start doling it out to patients - even during lead development.
What usually happens is that someone screens large collections of molecules for a specific activity - let's say the ability to stop vascular-cell derived cell lines from proliferating. Aha! We could use this compound to stop tumour angiogenesis and hence growth - maybe even starve a tumour to death. This compound becomes your new lead. There are still several problems to be overcome: it's quite toxic to other cell types too - turns out it binds a class of receptors, rather than a single one; it tends to kill mice into which you've injected it (at what you assume are) therapeutic doses - one of the breakdown products is really nasty; and it's virtually impossible to make on an industrial scale - it was produced by a third year grad student dicking around with some exotic ring structures.
So the first thing you need to do is figure out (approximately) how it works, and which part(s) of the molecule are active - if you can. Then you try to tweak the rest of it to change the undesirable properties - non-specific binding, breakdown products, solubility, etc. The only way to do this reliably is to use your chemist's intuition and all the organic theory you know to modify the compound, then test out the resultant properties. Again, and again, and again.
Bottom line is that we don't really know enough to predict these events in silico with any accuracy for a given unknown compound - yet. As rmc7777 mentioned, solving structures and some modelling is an outsourcable area, but we're a looooong way from fully-automated in silico drug design.
computational v. wet lab
That first paragraph in the original post certainly is ambiguous - I assume we are talking about a purely computational approach to e.g. rational drug design. My thoughts:
Are computational results useful to a drug designer
- yes
do we have to show some wetlab results to support the validity of these computational results
- yes
Computational biology is not a replacement for wet lab research. Ideally, it's a tool which complements your lab results (helps you make sense of the data) and directs your lab research (by rapidly identifying interesting new avenues). In the case of drug design, imagine something like molecular dynamics simulations of ligand/receptor binding using MPI on a large cluster. Given sufficient computational power, you could rapidly identify promising ligands and then test them in the lab, rather than randomly selecting compounds for lab trials.
So I'm not sure there's a market for pure computation in this field. I'd imagine most people who were serious about a "drug design startup" would consider both the computational and wet lab aspects. There might be opportunities in providing computational services to wet lab biologists who need them but have little experience.