Pseudoknot formation

One of the features of the model implemented in oxDNA is that it has a three-dimensional structure, so that it automatically incorporates topological effects (see Ref 3). In this example we will see how to initialise a single strand in with a sequence designed to form a pseudoknot in its final configuration.

The simplest option is to just generate a single strand with the right sequence, let it run at low temperature and wait for it to form the pseudoknot. Unfortunately, this can be very slow, as the formation of the pseudoknot requires two rare events (the closure of two hairpins). One should keep in mind that lowering the temperature to drive the formation of the pseuoknot can backfire, since the life of metastable states becomes longer and longer. Nevertheless, this is a perfectly acceptable solution.

A second option is to design a specific sequence, where we use "arbitrary" bases that only bind to one specific other base. This allows to eliminate metastable states in the pseudoknot formation, so that one can safely lower the temperature at will (keeping in mind that the model is physically meaningful only at temperatures where water is liquid).

A third option is to drive the formation of the pseudoknot with artificial forces, as one in the hairpin formation example. This is probably the most time-effective solution but requires a bit more work.

We study the following sequence that forms, at low enough temperature, drives the formation of a pesudoknot with two 6-base pair stems and two 8-base loops:

GTGCCGAAAAAAAACGCGAGACGGCACAAAAAAAACTCGCG

All the files needed to run this example are in the ${oxDNA}/EXAMPLES/PSEUDOKNOT directory. The following assumes you work in that directory.

Option 1

You can run this example directly in the OPT1/ directory.

First, we use the configuration generator to generate an input configuration for a single strand. The box size does not matter as long as it is big enough to contain the structure, so 30 s.u. will do. The file pkseq.txt contains the above sequence. So, run

../../UTILS/generate-sa.py 30. pkseq.txt

which generates generated.dat and generated.top.

Now, we choose to run a Brownian dynamics simulation since it is much more efficient than Monte Carlo (in our implementation) in exploring phase space. We use the standard "aggressive" values for the thermostat (see [Thermostat]). We use the average model in this example. We set a fairly low temperature, let's say 20 Celsius, significantly below the melting temperature of the two hairpins in the average-base representation. A long simulation might be needed, so we start with 10^9 total steps.

We can thus run

../../Release/oxDNA inputMD1

and wait (possibly a long time) until the pseudoknot forms. It is not guaranteed that the structure will for within the quite long simulation. To detect its formation, we can run the structure analyser to detect the formation of the correct bonds. Also, in the 5th column in the energy file (extensive base-pairing contribution to the total energy) you expect a number close to -8 for the formed structure (-0.7 times 12 base pairs). A single hairpin will have around half of that.

Option 2

You can run this example directly in the OPT2/ directory.

In this case, we have to repeat the all the steps above except that we need to manually tweak the topology file to have specific binding of the bases.

So, again generate the initial configuration with generate-sa.py and copy the topology file to a new one. The example directory already contains a working specific topology file for the impatient.

cp generated.dat specific.dat.

As discussed in the description of the topology file, specific bases can be set up with two-digit numbers in the second column of the topology file, and complementarity is implemented if the sum is equal to 3 (negative numbers can be used). So if we want base number 0 and 26 to be bound, we can set their "type" (second column) to be 100 and -97, respectively. Pay close attention when modifying by hand the topology file, since it is very easy to make mistakes (base types don't add up to 3, using the wrong number, etc.). It makes sense to automatise this task.

In this case, we want bases 0-5 to be bound to bases 26-21 and bases 14-19 to be bound to bases 41-36. Modify the topology file so that complementary pairs have types corresponding to numbers with magnitude grater than 10 and that add up to 3. The poly-A sections in the loop can be left as A's in the topology file.

If you look into the file inputMD2, you will see that it is the same as inputMD1 except that it uses specific.top as the topology file. If you decide to use the ready specific topology file, just copy it to specific.top. The temperature has also been lowered to 0C, since we expect no unwanted metastable states.

You can now run the code and see what happens. It should make base-pairs more rarely, but it should make only correct ones.

../../Release/oxDNA inputMD2

Option 3

Why wait if we just want to form a structure and are not interested in how it forms? The mutual_trap force is implemented exactly to form initial configurations, having no physical meaning. But getting an initial configuration is sometimes half of the job.

In this case we can use either of the topology file described above. We choose to use Monte Carlo, though, so that we don't have to worry about tweaking the magnitude of the traps and the thermostat parameters not to make the system explode.

First of all, we have to generate a file where we specify the external forces. We set up 4 traps, a mutual trap between particles 0 and 26 and a mutual trap between particles 14 and 39. These should be enough to drive the formation of the two stems in a relatively short time.

So, for each of the pairs, we set up a trap like this in external.conf:

{

type = mutual_trap

particle = 0

ref_particle = 26

stiff = 1.

r0 = 1.5

}


{

type = mutual_trap

particle = 0

ref_particle = 26

stiff = 1.

r0 = 1.5

}

Repeat the same block with the indexes 0 and 26 changed to 14 and 39.

The file inputMC is a Monte Carlo input file with everything set up to use the external.conf file you just created.

Running the program

../../Release/oxDNA inputMC

should produce a pseudoknot within half an hour, maybe faster. In this case we don't need the temperature to drive the formation of the motif, so we can use room temperature.

If you have any questions about this example, you can e-mail Flavio Romano.