Hacker News Reader

Comment by jballanc | Hacker News Reader

There's a very interesting rule of thumb that is used in the protein folding field. If 30% of the amino acids in two different proteins are identical, then you can be very certain that they will have identical (or nearly identical) 3D shapes. However, it is entirely possible that two proteins have no sequence in common, yet still fold to the same shape. A friend actually stumbled upon just such a case when I was still doing protein folding work, and it was amazing to look at the sequence of these two different proteins and see nothing in common, then flip to the 3D structure and see that they line up almost perfectly.

Proteins are immensely cool.

View on HN

That's why ab initio protein structure prediction is such a sought-after tool in bioinformatics. You can have an incredibly high degree of nucleotide/amino acid substitution between two proteins with the same structure, function and evolutionary history, but current methods of detecting these relationships still largely rely on sequence comparison alone.

I saw Rhiju Das speak last week (really fantastic talk). He's working on some very cool deterministic ab initio folding approaches. http://www.stanford.edu/~rhiju/research.html

The last time I read into this subject, there were neural network models that achieved around 80% accuracy. There's a research group at my uni dedicated to this research area.

I think you're referring to the prediction of "secondary structure" of proteins, which was attempted using neural network models as early as 1989. [1] Predicting secondary structure is child's play compared to predicting the folded tertiary structure of a protein, hence the continued use of insanely computationally expensive ab initio folding efforts like Folding@Home.

[1] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC286422/

Indeed! And crazier still, you can have two proteins with substantial structural homology, but with entirely different functions and mechanisms.

When I worked in structural bio, one of my colleagues was studying a protein [complex] that had a prominent structure homologous to a helicase, but no DNA-unwinding activity whatsoever. To this day it's unknown what that structure does.

I'm excited to more progress linking structure to function, even if this is very very difficult.

I have no idea what "protein folding" does, but you just sold this story so well, that I'm about to look it up.

My point: great comment, makes me want to learn.

protein folding is the process by which a sequence of amino acids - a protein - assumes a 3d structure, made up of "local" units of structures, often turns, sheets (beta sheets) and cylinders (alpha helixes) further arranged into a larger super structure. this is what confers functionality (binding, enzymatic reactions, etc) and dynamics (movements integral to the protein's functionality). the chemical principles that drive this are things like hydrophobic amino acids burying themselves away from a typical aqueous (water-based, or at least highly polar) environment, ionic pairings, hydrogen bonding (strong but not permanent bonds), etc.

in terms of predicting the protein's structure, the challenge is the sheer number of computations, the dynamics the protein goes through, and the effects of any environmental factors as the protein is synthesized or folded.

i spent much of a decade (mid 90's to early 00s) studying folding with an aim to getting into enzyme engineering. fun stuff, but i have since left biochem.

Reply on news.ycombinator.com