Other Notes

Scoring on Less Metrics

Normally, four metrics are used to score each origami scaffold sequence (mode 2 operation of scaffoldselector), and three metrics are used (excluding metric 2) when the scaffold strand is a fixed sequence and it is just being rotated (mode 1 operation of scaffoldselector).

These defaults can be changed, by modifying the "mode 1 metrics" and "mode 2 metrics" settings in the settings.json file. Scoring with less metrics takes less computation time but changes the dimension of the pareto front of optimal scaffold sequence solutions, leading to different top-ranked scaffold sequences.

Run Results with Less Metrics

When all origami scaffold sequences have been scored already, the results analysis can be run again using less metrics. For example, using:

python3 selector.py myorigami results -M12

will produce the results HTML analysis using only metrics 1 and 2 (leaving out metrics 3 and 4 from the pareto front calculations). At least two metrics must be specified. This operation is useful because it allows to see how changing the dimensionality of sequence selection changes the outcome, without having the re-score all origami scaffolds.

DNA Origami Recommended

It is recommended to use scaffoldselector with DNA origami, and set in the scaffold selector settings.json file:

"scaffold material"                    : "DNA",
"staple material"                      : "DNA"

The energy model used is determined by the scaffold material. The scaffoldselector also has a basic RNA energy model which is used whenever "scaffold material" is set to "RNA". In this mode G-U wobble pairs are also included. However, the RNA energy model is not recommended, as it is currently less accurate than the DNA energy model at identifying RNA-RNA binding sites and requires further development work.

If the DNA energy model is used, with RNA staples, e.g:

"scaffold material"                    : "DNA",
"staple material"                      : "RNA"

then G-U wobble pairs are also counted as valid base pairs (of same strength as the standard Watson-Crick base pairs) when it comes to determining binding sites.

Code Optimisation

Currently all code is Python 3. In future, Metrics 1 and 2 could benefit from code optimisations (like JIT compilation to machine code) to increase their calculation speed. However, Metrics 3 and 4 are calculated via calling the NUPACK Python library and are thus limited by the speed of the NUPACK library implementation.