Will The Need For Certain Bioinformatics Approaches Be Eliminated By Advances In Microscopy?

Question

[Nathan Nehrt]

I've often wondered if (or when) someone will create a microscope powerful enough to track an individual coding gene from transcription to splicing to translation, then on to watching the protein interact with other proteins in the cell or to perform other functions. Assuming that it would be possible to identify the individual genes and proteins being tracked, would this eliminate the need for computational approaches to predict protein interactions or to predict GO term annotations? Or, if individual post-translational modifications could be identified, would this eliminate the need for computational prediction of PTMs?

Of course, these approaches would require a revolutionary leap forward in microscopy, and would likely still be limited if the process can only be observed in vitro (out of the normal biological context). But, assuming such a leap forward could be made, which other bioinformatics approaches could potentially be made obsolete?

Answer 1

[Eric Fournier]

Techniques for "observing" protein-protein/gene-proteins interactions already exist. NMR and x-ray crystalography do exactly that: get us 3D structures of molecules of interest. They may not be traditional microscopy, but they accomplish the same goals.

Disregarding that, I doubt any traditional microscopy approach, as described, could ever provide such results. This is a question of scope:

1. First off, these phenomena happen so fast and on such a small scale it would be practically impossible to focus on a single molecule and "catch" the events of interest.
2. Secondly, the kinetics of most such systems require extremely large number of molecules to occur in a biologically significant manner. You can't just put in one transcript and one transcription factor in a vial and hope that brownian movement will bring them close enough together to interact.
3. Thirdly, even if you could somehow manage to directly observe, say, a gene-transcription factor interaction, if you consider the number of genes x the number of transcription factors, you end up with millions of possible combinations, which cannot possibly all be checked out individually.

[Nathan Nehrt]

Hi Eric - yes, it is definitely questionable whether microscopy will ever be able to do this. There would be many hurdles to overcome. I was thinking of the high false positive rate with many computational approaches to predicting protein interactions, and was thinking more "wouldn't it be great if" such a thing was possible. What if we could see inside cells like "The Inner Life of the Cell" (http://www.youtube.com/watch?v=Mszlckmc4Hw) How would this affect bioinformatics? But, no, I don't think it's happening any time soon.

[Eric Fournier]

My point was that if we could indeed see in cells like in "The Inner Life of the Cell", what we would see would be the equivalent of taking all of the "Where's Waldo" books in existence, superposing them, and then animating them.

It would be such a terrible mess that nobody could ever make any sense out of it.

[Nathan Nehrt]

Yes - I was surprised by the image depicting the densely packed E.coli cytoplasm shown here. Tracking a single protein in this cell would be extremely difficult.

Answer 2

[Chris Evelo]

Your leap forward in microscopy, making it able to see individual bases, would not only require a leap forward microscopy but also a radical change in the physics underlying microscopy. In other words it will never happen. Check for instance: http://www.ou.edu/research/electron/bmz5364/resolutn.html. Or any physics book explaining about the uncertainty principles. I think the subjective tag is not even necessary here, this is objectively never going to work.

[Nathan Nehrt]

Hi Chris - point taken. I probably should have tagged my question as "Friday morning daydreaming".

[Chris Evelo]

Sorry if that was too harsh. Daydreaming can be fun.

[Michael Schubert]

+1 for resolution; basically, shorter wavelength = better resolution but also more energy, which tends to destroy samples (or just go right through them)

Answer 3

[Casey Bergman]

Assuming such technologies could be developed, the answer to your question is yes. But such a leap forward would consequently create new bioinformatics challenges in the field of image bioinformatics. Current advances in imaging already require new methods for stitching 2D and 3D images of large objects that are bigger than one field of view and for automatic comparison of images. Combining and comparing 4D images from live samples will present even further bioinformatic challenges, so even if advanced imaging kills off some areas of bioinformatics that are dependent on current technology, it will most likely create new areas of bioinformatics. Viva bioinformatics!