ASU model (Figures 8 & 9)

At the time of our publication the 2017 ASU iGEM team did not have all of their parts (with the subparts) entered into the Registry of Standard Biological Parts. We began manually building the model by referring to their wiki page. After contacting the team to get a more accurate description of each part we were given access to a collection of their parts on a cloud software service called Benchling. We translated the Benchling data into Excel format found here. We used this information to create a feature model representing the ASU experiments. The complete Feature Model represents 90 products, and the ASU team ran a subset of 30 product experiments by adding constraints.

As of publishing our artifacts, ASU has both published their results [1] and made the collection of their parts public (link).

Protein Slice (Figure 11)

If the ASU team wanted to focus on the proteins, which function as the main communication between the sender, receiver, and reporter, they could slice the cell-to-cell signaling model. We use the slice feature of FeatureIDE to get this sliced model. This slice is already included in the FeatureIDE workspace we provide. To generate your own slice:

• • Find the Cell_to_Cell_Signaling model (or another model of your choice)

  • Right-click on the .xml model file

  • Under FeatureIDE select Slice Feature Model

  • Select the features you want to include in your slice

Our Protein Slice creates 100 products. This is significantly fewer than the total space (7.50169 × 10²⁰), but more than what the ASU team tested (30).

Covering Arrays (CIT Samples)

We have two 2-way covering arrays in this work, one for the ASU model and one for the cell-to-cell signaling model. To generate the samples we used the CASA tool [2]. Further information on CASA can be found here. CASA source files can be directly downloaded here.

Due to the simplicity of the ASU model (no constraints and high commonality) we create the sample on the three features that vary (10 proteins of the sender, 3 proteins of the regulator, and 3 proteins of the activation of the behavior). This generates a sample with 30 tests covering a total of 69 pairs of the features.

The cell-to-cell signaling model was more complicated, for example the cardinality of the terminators (choose 1 or 2 terminators) had to be encoded. To allow the promoter to choose 1, 2, or 3 of its subfeatures, constraints were implemented. This generated a sample with 741 tests, and with constraints

All covering array inputs and outputs can be downloaded here.

Reverse-engineered Model (Figure 13)

In this paper, we reverse-engineered the ASU model from 30 products using SPLRevO [3]. We provide this model (ASU-r03) under the models directory of the download package. We configured our genetic algorithm to use FFValidity fitness function and 1% mutation on a population of 100 models. We ran the model up to 200 generations on the computing cluster with AMD Opteron(TM) CPUs running at 2300MHz with a maximum Java memory pool of 32GB. Then, we captured the model with the best fitness value (maximization). The results are in the results directory. It shows that SPLRevO was able to provide us with a model that closely resembles the hand-built model and has 100% validity (it represents exactly the same number of products).

The existing tool, SPLRevO is limited in scalability when reverse-engineering from products (limits to 27 features) so we slightly condensed the ASU model in two ways. We (1) combined any double terminator (B0010 and B0012) into one (B0015). This is the case for the sender, receiver, and behavior (e.g. B0010_S and B0012_S become B0015_S). BioBrick part B0015 is the combination of parts B0010 and B0012. And (2) removed two features mandatory to all products (B0034_S_mC and mCherry). Note this still results in 30 products.

The tool and instructions on how to run on our models can be downloaded here.

You can either run the experiment yourself with the model in the /models/ directory or view our outputs in the /models/sample_results/ directory. Please note for the ASU model, we ran SPLRevO for about 27 hours with 13 hours of CPU time to get the correct model. We use the result of the 100% validity model (available here) for this work (see paper discussion and SPLRevO documentation for more). We translated this output by hand into FeatureIDE format. The key for the SPLRevO format can be seen in the Table below [4]. The translated model can be seen in the feature model below and in the Eclipse FeatureIDE workspace provided in these artifacts.