[Market strategy] Even energy distribution in design (neckarea excluded)

I would like to ad a strategy that says, that energy difference between strings of same length must be highest 5%.

Eg. Branches designs have:

4 arms which is four nucleotides long
2 arms that are five nucleotides long

The energy difference between arms of same lenght must be max 5% difference between the arm with lowest collective energy and the arm with highest collective energy.

Penalize with -1 for each 1% higher than 5 % energy difference between arms of equal size.

And ad up for different lenghts of arms:
4 arms, 4 nucleotides long - 8% difference (3% over 5%)
2 arms, 5 nucleotides long - 7% difference (2% over 5%)

3% + 2% = 5% Penalize with -5

Dear Eli,

Your strategy has been added to our implementation queue with task id 37. You can check the schedule of the implementation here.

ETA of the implementation is 7/12/2011

Thanks for sharing your idea!

EteRNA team

Hi Eli,

Can you explain how “arms” are defined? In general, we need all terms with specific definitions to have algorithms implemented.


Hi Jee

Ah, I see where the problem is. With arms I did not mean the whole arm in eg. a four armed design. Here I meant string/stem. Where RNA is double stranded.

I will like to have the tolerance for energy even energy difference set higher. To a max 20-25 % difference between the string with the lowest energy concentration (compared to length of string) to the string with the highest energy concentration. I think I got it set too low before and having it too low will exclude many good designs. Even energy distribution seem to be the rule for good designs, but it is a more extreme energy difference between parts of a RNA design that will make a design do bad.

I think there is a local part of this strategy, that should ensure that local elements like strings don’t have uneven distributed energy, like a 4 nt long string having either all GC-pairs or all AU-pairs.

Then there is a global part of it. But here is a bit more about how the global part of it could be made, if I look at the energies in the quad. Quads have energies between -0.9 to -3.4. That is a difference on 2.5. If I divide that with 4, and add a fourth of that to both the low energy quad and the high energy quad, I have an energy area between -1.53 to -2.8 which is relatively safe for a design to be in. And a bottom and top energy area that I don’t want the designs to be in.

If a good mean for global energy distribution could be established, perhaps it could be used to catch energy violations. Imagine that the mean energy for an element like a string is -3.4 (christmas tree) or -0.9 (cub scout). Then the difference in good global energy for a design (can be a number or an interval of numbers, to ease calculation for our bot) can reveal that the local mean is way off.

And as I see I didn’t put in a link to the theory behind this strategy, when I made it, here it comes. Theory of even energy distribution - Energy, structure and symmetric colors