In the January 2014 issue of Scientific American is an interesting article about simulating a cell in software.
It seems that scientists are getting pretty good at it.
A researcher who has done a lot in the field, Markus Covert, writes about their most recent attempt at this. Unlike many previous attempts, this group concentrated on a single cell, and they used the most simple cell, Mycoplasma genitalium. Scouring the literature for anything concerning the structure, behaviour, energetics and biochemistry of the organism, they built their simulator over a number of years, and now, given that it works so well, find that it can inform them on things about the organism in such a manner that:-
1) they learn new things, and
2) can direct their research to verify things discovered or suggested by simulator runs.
The Sci Am article is referenced here:-
Scientists Successfully Model a Living Cell with Software
An online research report describing the work is here:-
A whole-cell computational model predicts phenotype from genotype
The Sci Am article was fascinating, although a read of the research article might end up being more rewarding.
Among other things, what I found fascinating were the following:-
Underlying random processes
As an example of this, proteins don’t necessarily bind to DNA securely. The are often getting knocked off their regulatory regions by the random battering from other proteins. This can happen up to 30,000 times per 9 hour cell cycle. The researchers were surprised to see this.
Emergent phenomena
The particular cell they simulated has a cell cycle period, time between cell divisions, of around 9 hours. Their software cell also divides at that time. Interestingly, this proved to be an emergent phenomenon caused by “the interaction of two distinct phases of replication, each of which varies wildly in duration”.
Anyway, for those who are interested - it’s a good read.
It seems that scientists are getting pretty good at it.
A researcher who has done a lot in the field, Markus Covert, writes about their most recent attempt at this. Unlike many previous attempts, this group concentrated on a single cell, and they used the most simple cell, Mycoplasma genitalium. Scouring the literature for anything concerning the structure, behaviour, energetics and biochemistry of the organism, they built their simulator over a number of years, and now, given that it works so well, find that it can inform them on things about the organism in such a manner that:-
1) they learn new things, and
2) can direct their research to verify things discovered or suggested by simulator runs.
The Sci Am article is referenced here:-
Scientists Successfully Model a Living Cell with Software
An online research report describing the work is here:-
A whole-cell computational model predicts phenotype from genotype
The Sci Am article was fascinating, although a read of the research article might end up being more rewarding.
Among other things, what I found fascinating were the following:-
Underlying random processes
As an example of this, proteins don’t necessarily bind to DNA securely. The are often getting knocked off their regulatory regions by the random battering from other proteins. This can happen up to 30,000 times per 9 hour cell cycle. The researchers were surprised to see this.
Emergent phenomena
The particular cell they simulated has a cell cycle period, time between cell divisions, of around 9 hours. Their software cell also divides at that time. Interestingly, this proved to be an emergent phenomenon caused by “the interaction of two distinct phases of replication, each of which varies wildly in duration”.
Anyway, for those who are interested - it’s a good read.
Comment