Team:Virginia/Model
Loading
#3f5eae
Modeling
Index:
Protein Structure Modeling
Fusion proteins in Manifold
Zinc finger fusion proteins provide the "glue" needed to make Manifold come together. Zinc fingers are small proteins that bind to certain DNA motifs with high precision. They can be "fused" to other proteins by appending their genetic code to one end of another protein's gene. Thus, other proteins can be given zinc fingers' ability to localize to DNA motifs. In Manifold, the biosynthetic pathway enzymes are translated as fusion proteins with the zinc fingers ZFa and ZFb, causing them to attach to the DNA scaffolds within our bacterial microcompartments (BMCs) with desired order and spacing. Additionally, DNA scaffolds are adhered to the BMCs by a fusion of the zinc finger ZFc and the N-terminus of PduD, a BMC proteins that naturally binds to the interior of the wall protein PduA. While these fusion proteins give Manifold the its ability to assemble multiple moving parts into a functioning unit, they also pose a risk; adding a zinc finger to a protein could cause it to fold poorly in vivo, inhibiting the proteins' functional domains. To utilize fusion proteins while preventing the misfolding of the enzymes that are needed to actually make resveratrol, we used protein structure prediction software to analyze the conservation of fusion proteins' structures.
From I-TASSER to AlphaFold
We started by modeling all proteins involved in Manifold with I-TASSER, the former leading protein structure prediction tool. After running initial job's on the Yang Zhang lab's remote webserver, we transitioned to installing I-TASSER on the University of Virginia's Rivanna supercomputer, where nearly 10,000 hours of CPU time were exhausted generating structures of proteins and their fusions with zinc fingers. After obtaining initial results, we sought to compare alternative fusion protein designs. Assembling Manifold required making three key decisions:
- Acetyl-CoA synthetase (ACS) fusion: ACS with ZFa attached at either N- or C-terminus
- Acetyl-CoA carboxylase (ACC) fusion: Either ACC-α or ACC-β subunit with ZFb attached at either N- or C-terminus
- PduD fusion: Full PduD, first 146 amino acids of PduD, or first 23 amino acids of PduD with ZFc attached at C-terminus
While initial I-TASSER results were largely inconclusive due to the sheer computational complexity of predicting large protein structures, the surprise release of DeepMind's AlphaFold2 on July 15th gave new life to this modeling project. AlphaFold uses artificial intelligence to go beyond the capabilities of more deterministic methods like I-TASSER. It achieved a median CASP score of 92.4 GDT
[1]The AlphaFold team. (2021). AlphaFold: a solution to a 50-year-old grand challenge in biology. DeepMind. https://deepmind.com/blog/article/alphafold-a-solution-to-a-50-year-old-grand-challenge-in-biology
, significantly higher than any previous software. After getting AlphaFold set up on Rivanna, we were able to redo our work with I-TASSER, this time with vastly accelerated runtimes from GPU cluster support and superior results.
Assumptions of Structural Modeling
While AlphaFold results often look convincingly realistic, it is important to remember the assumptions made when using it to predict protein structure. Primarily, AlphaFold does not take into consideration external influences that are influential in vivo, like the effect of oxidizing cytosol on disulfide bond formation
[2]Darren Fast. (1995). DISULFIDE BONDS. Birbeck, University of London. http://www.cryst.bbk.ac.uk/PPS95/us/darren-f/s-s.html
. Due to computational limitations, AlphaFold treats each protein as if it exists in isolation, submerged in a generic solution.
Fig 1. Left: Molecule created by predicting structure of PduB' monomer with AlphaFold and then creating rotated copies in ProtKit to modelnatural Pdu homohexamer. Imaged with NGL Viewer.
[3]AS Rose and PW Hildebrand. NGL Viewer: a web application for molecular visualization. Nucl Acids Res (1 July 2015) 43 (W1): W576-W579 first published online April 29, 2015. doi:10.1093/nar/gkv402
[4]AS Rose, AR Bradley, Y Valasatava, JM Duarte, A Prlić and PW Rose. Web-based molecular graphics for large complexes. ACM Proceedings of the 21st International Conference on Web3D Technology (Web3D '16): 185-186, 2016. doi:10.1145/2945292.2945324
Right: PduB homohexamer from RCSB Protein Data Bank obtained via 1.64 Ã… X-ray diffraction crystallography. The similarity of the two models showcases AlphaFold's extraordinary predictive accuracy, even when modeling a single piece of an oligomeric protein with no contextual information about the folding environment. [5]Pang, A.H., Pickersgill, R.W. (2014). 4I61:
Crystal structure of a trimeric bacterial microcompartment shell protein PduB. Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB). https://www.rcsb.org/structure/4I61
Additionally, our usage of AlphaFold failed to account for the oligomeric quaternary structure of many proteins in Manifold. The Pdu BMC's wall proteins, like PduA and PduU, form homohexameric complexes that tesselate together
[6]Crowley, C.S., Sawaya, M.R., Yeates, T.O. (2008). 3CGI:
Crystal structure of the PduU shell protein from the Pdu microcompartment. Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB). https://www.rcsb.org/structure/3CGI
. While we were still able to establish confidence in our more impactful results, we noticed AlphaFold's inability to predict the structure of the PduK monomer; AlphaFold would invariably predict it to have an unfeasibly large circular loop of random coil. This issue potentially corroborates suspicion of the significance of PduK in the Pdu operon that was discussed in our correspondence with the Yeates Lab at UCLA.
Fig 2. AlphaFold structure prediction for PduK monomer. The large circular loop of random coil (regions lacking of secondary structure) could never occur in nature, indicating a flaw in AlphaFold's performance.
Structural Modeling Conclusions
By comparing AlphaFold results for different configurations of the same fusion enzymes for quality of structural preservation compared to the original enzyme, we were able to conclude that use of ACS with the N-terminal zinc finger and ACC-βwith the N-terminal zinc finger. Additionally, we chose to connect our scaffolds to the BMC walls using the first 23 amino acids of the N-terminus of PduD, which contains its N-terminal helix evolved to intertwine with the a helix on the interior of PduA wall proteins.
Fig 3. Left: Cross-sectional model of fusion protein of N-terminal helix of PduD and ZFc bound to PduA homohexamer. Center: ACS with ZFa fused at its N-terminus. ZFa at bottom of figure. Right: ACS with ZFa fused at its C-terminus. ZFa at bottom of figure. Note the superior organization of the center protein versus the left protein. All images created with ProtKit, NGLViewer, and the RCSB Pairwise Structure Alignment Tool.
Molecular Docking
Docking Simulations
In order to predict whether our fusion enzymes would still be functional, molecular docking simulations were performed using AutoDock Vina and the HDOCK server. Fusion proteins are influenced by not only the proteins make up the fusion but also by the design of their linker which can influence the fusion in a number of different ways including effects such as increased stability, expression, and activity
[7]Chen, X., Zaro, J., & Shen, W.-C. (2013). Fusion Protein Linkers: Property, Design and Functionality. Advanced Drug Delivery Reviews, 65(10), 1357–1369. https://doi.org/10.1016/j.addr.2012.09.039
. While the design of the linkers was determined through literature review and known examples, variations regarding which terminus should be connected to the linker still needed to be considered. Depending on where the linker is located, this can impact the binding affinity of not only the enzymes to their ligands but also the binding affinity of the fusions towards the DNA scaffolds as well as their spatial orientation.These factors are critical to the spatial channeling effect of the manifold system, as such we sought to predict these interactions to the best of our ability using the tools best suited for this task.
AutoDock Vina
Differences in N-terminal or C-terminal linkers can result in differences in our enzymes’ binding affinities. Depending on where the zinc finger is located it can noncovalently interact with our ligands, thereby decreasing enzymatic activity. Ideally, the zinc finger protein should be located as far away as possible from the active sites on our enzymes which should have high binding affinity towards their ligands, however the location of these active sites are unknown in literature. Thus once predicted fusion proteins were generated from our protein folding efforts, which confirmed the proper folding of these proteins, we then sought to answer where the linker between the fused proteins should be located to minimize the possibility of the zinc finger interacting with our ligands. Determination of these factors required the use of virtual screening tools, of which we sought to answer the following questions:
Questions
- Where will our ligands interact and dock on our enzymes during the biosynthesis of resveratrol and what is the energy associated with this?
- Which fusion protein variations would maximize the binding affinity of our ligands to our enzymes whilst also minimizing the interactions with the zinc finger protein?
To resolve these questions, we used a common virtual screening tool known as AutoDock Vina
[8]Trott, O., & Olson, A. J. (2010). AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization and multithreading. Journal of Computational Chemistry, 31(2), 455–461. https://doi.org/10.1002/jcc.21334
. Vina is a molecular docking simulation that is based on a gradient-optimum conformational search that provides a series of stochastics conformations. While the software performs these calculations rapidly, it is not without its faults. Autodock Vina has been found to be more accurate when predicting larger targets with charged binding locations as compared to AutoDock [9]Vieira, T. F., & Sousa, S. F. (2019). Comparing AutoDock and Vina in Ligand/Decoy Discrimination for Virtual Screening. Applied Sciences, 9(21), 4538. https://doi.org/10.3390/app9214538
. AutoDock Vina 1.1.2 also lacks the ability to consider explicit hydration, which can be critical in mediating receptor ligand interactions. Furthermore, AutoDock Vina by default assumes that the receptor will dock with the ligand rigidly and while it is possible to “split†the receptor into flexible and rigid components, this requires the use of additional tools and may not fully capture the degree of flexibility that the receptor can exhibit. As such several assumptions were made during the generation of our docking simulations:
Assumptions
- Conformations between the receptor and ligand will be rigid.
- Docking between the receptor and ligand will occur without hydration.
Given the lack of information regarding the active sites on the enzymes ACC and ACS, blind docking was performed on both fusion and nonfusion enzymes. This generated several possible locations where the ligands could bind to the protein and their respective binding affinity. Based on these results, we then chose our fusion enzymes to be ACS, ACCα, and ACCβ with N-terminal zinc finger proteins due to their clear and close binding with their ligands.
Fig 4. AutoDock Vina prediction of Acetyl-CoA synthetase N-terminal Zinc Finger A fusion docking with CoA and Acetate
HDOCK Server
Differences in N-terminal or C-terminal linkers can also result in differences in the ability of our zinc finger proteins to bind to their binding motifs on the DNA scaffold. As noted before, proteins of a fusion protein can interact with each other, potentially impacting their behavior. Thus another aspect that we considered was the binding affinity of our fusion proteins towards their respective zinc finger binding site. Successfully binding of our fusion enzymes’ zinc finger to their binding site on the DNA scaffold is critical for the spatial channeling of the resveratrol biosynthesis pathway, which is a major component of the Manifold system. Therefore, our ideal fusion protein would exhibit high binding affinity towards its motif as well as binding in an orientation that would allow for spatially unimpeded enzymatic activity. Given our successful protein folding efforts, we used the generated protein structures for virtual screening with their zinc finger binding motifs. Thus, our approach to describe these factors followed a similar path as what we did with AutoDock Vina but with different tools and the following questions:
Questions
- What will be the binding affinity between our fusion proteins’ zinc finger and it's DNA motif?
- Which fusion variation would maximize the binding affinity of our fusion protein’s zinc finger and not spatially constrain the active sites on the enzyme?
To describe these questions, we used the HDOCK server for DNA and protein docking. HDOCK is a public web server that utilizes its own docking algorithm and several different scoring functions to predict the conformations of a receptor and its ligand
[10]Yan, Y., Zhang, D., Zhou, P., Li, B., & Huang, S.-Y. (2017). HDOCK: A web server for protein–protein and protein–DNA/RNA docking based on a hybrid strategy. Nucleic Acids Research, 45(Web Server issue), W365. https://doi.org/10.1093/nar/gkx407
. HDOCK is unique amongst many other similar docking servers for its capability to perform protein-protein and protein-DNA/RNA docking using different scoring functions to accurately predict these interactions [11]Huang, S.-Y., & Zou, X. (2008). An iterative knowledge-based scoring function for protein-protein recognition. Proteins, 72(2), 557–579. https://doi.org/10.1002/prot.21949
[12]Huang, S.-Y., & Zou, X. (2014). A knowledge-based scoring function for protein-RNA interactions derived from a statistical mechanics-based iterative method. Nucleic Acids Research, 42(7), e55. https://doi.org/10.1093/nar/gku077
. In the case of protein and DNA/RNA docking, due to the flexible nature of nucleic acid sequences, prediction of docking can be difficult due to the numerous conformations that DNA could become and the numerous interactions that a protein could have with these conformations. HDOCK utilizes an iterative statistical mechanics based method such that possible conformations are examined and recorded. Each of these possibilities are then scored through a method known as ITScore-PR that considers the potential or energy derived from the distance of each atom in the protein-DNA conformation. The use of this scoring method is also key for implicitly taking into account the effect of DNA flexibility as current docking algorithms are not yet able to accurately handle this effect. ITScore-PR does this by applying an energy penalty at specific distances between atoms. In total, these methods have made HDOCK among the top docking algorithms according to CARPI.
From our use of the HDOCK servers, we were able to receive several potential conformations that our fusions would make with their DNA motifs and the affinities associated with them. Based on these results, we were able to further verify the succesful docking of our chosen enzyme fusion proteins with their respective zinc finger binding motifs.
Fig 5. Left: HDOCK Prediction of Acetyl-CoA carboxylase α Zinc Finger B fusion with DNA Scaffold. Right: HDOCK Prediction of Acetyl-CoA carboxylase β Zinc Finger B fusion with DNA Scaffold.
Molecular Dynamics Simulation
What is Molecular Dynamics Simulation
An important aspect of the Manifold System is the ability for reactants and products of the biosynthetic pathway to be able to pass through the BMC pores. This is essential for continuous production, as new reactants can enter the BMC “bioreactor†and finished products leave. Determining whether the BMC pores are permeable to specific molecules is crucial for evaluating the compatibility that the Manifold system has with a biosynthesis pathway, however this is a difficult question. Different biosynthesis pathways can have different chassis and molecules involved and likewise these molecules will have different permeabilities through the BMC pore. Thus, we decided to use molecular dynamics simulations to determine these values due to the flexibility these simulations have in regards to input.
To describe the transport of our ligands through the BMC pore, we utilized version 2021.3 of GROMACS, an MDS software which uses statistical mechanics and Newtonian motion to describe complex chemical systems atomistically
[13]Abraham, M. J., Murtola, T., Schulz, R., Páll, S., Smith, J. C., Hess, B., & Lindahl, E. (2015). GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX, 1–2, 19–25.
. MD simulations work by simulating every atom in a system as hard spheres with a volume determined by their Van Der Waal potentials and set mass. Bonds between atoms are represented as their spherical volumes combining together and can either behave as stiff or stretchy depending on the simulation. Following simulation setup and other settings, the atoms within the system are then assigned a range of velocities at random based on a Boltzman distribution on temperature. As the simulation proceeds, Newtonian physics is applied based on the forcefield of the MD simulation and each atom is then able to interact with each other, transferring energy and momentum by their electrostatic and Van Der Waals interactions. Based on this, MD simulations offer a way to predict the macro behaviors of systems, such as their stability, conformations, and energy.
Simulation Setup
Key to generating quality MD simulations is the proper establishment of our system. This includes setting up proper conditions, such as ion concentration and temperature, to accurately reflect intracellular conditions within an E Coli as well as obtaining structural files of our molecules and proteins of interest. Parameters of cellular conditions were obtained through literature and are the following:
Table of Parameters
| Parameter | Value |
|---|---|
| Intracellular pH | 7.2-7.8 (7.5) [14]Keith A. Martinez, I. I. (2012). Cytoplasmic pH Response to Acid Stress in Individual Cells of Escherichia coli and Bacillus subtilis Observed by Fluorescence Ratio Imaging Microscopy. Applied and Environmental Microbiology, 78(10), 3706. https://doi.org/10.1128/AEM.00354-12 |
| Temperature | 300 K [15]Kumar, P., & Libchaber, A. (2013). Pressure and Temperature Dependence of Growth and Morphology of Escherichia coli: Experiments and Stochastic Model. Biophysical Journal, 105(3), 783. https://doi.org/10.1016/j.bpj.2013.06.029 |
| Concentration of NaCl | 5 mM [16]Shabala, L., Bowman, J., Brown, J., Ross, T., McMeekin, T., & Shabala, S. (2009). Ion transport and osmotic adjustment in Escherichia coli in response to ionic and non-ionic osmotica. Environmental Microbiology, 11(1), 137–148. https://doi.org/10.1111/j.1462-2920.2008.01748.x |
Structural files were obtained from our protein folding efforts as well as through online servers such as PRODRG which performs the proper modifications needed to make a molecule compatible with GROMACS force fields
[17]Schüttelkopf, A. W., & van Aalten, D. M. F. (2004). PRODRG: A tool for high-throughput crystallography of protein-ligand complexes. Acta Crystallographica. Section D, Biological Crystallography, 60(Pt 8), 1355–1363. https://doi.org/10.1107/S0907444904011679
. Protein protonation is also a critical aspect of setting up our simulation as protonation states will affect the charge and therefore interactions a protein will have with other atoms. To determine the protonation state of our atoms we used web servers such as the H++ server which predicts these states on each of our amino acid residues [18]Ramu Anandakrishnan, Boris Aguilar and Alexey V. Onufriev, "H++ 3.0: automating pK prediction and the preparation of biomolecular structures for atomistic molecular modeling and simulation", Nucleic Acids Res., 40(W1):W537-541. (2012)
[19]Myers, J., Grothaus, G., Narayanan, S., & Onufriev, A. (2006). A simple clustering algorithm can be accurate enough for use in calculations of pKs in macromolecules. Proteins: Structure, Function, and Bioinformatics, 63(4), 928–938. https://doi.org/10.1002/prot.20922
[20]Gordon JC, Myers JB, Folta T, Shoja V, Heath LS and Onufriev A., "H++: a server for estimating pKas and adding missing hydrogens to macromolecules", Nucleic Acids Res. Jul 1;33:W368-71. (2005).
. Once these parameters were determined, we then ran our simulation through energy minimization, NVT, and then NPT ensembles to further equilibrate the system. NVT ensemble runs the system with constant number of particles, volume, and temperature and NPT ensemble runs the system with constant number of particles, pressure, and temperature. The combination of all these ensembles with differing settings applied for each equilibration ensure that our simulation will not blow up.
Umbrella Sampling
Following the generation of our stable equilibrated system, we then performed a sampling technique known as umbrella sampling on our system
[21]Liao, Q. (2020). Chapter Four—Enhanced sampling and free energy calculations for protein simulations. In B. Strodel & B. Barz (Eds.), Progress in Molecular Biology and Translational Science (Vol. 170, pp. 177–213). Academic Press. https://doi.org/10.1016/bs.pmbts.2020.01.006
. Umbrella sampling consists of introducing a pulling force onto a target molecule to generate the initial configurations of that molecule along a certain coordinate. From this path, “windows†or states of where that molecule was during the initial pulling are selected at specific intervals. Each window is then treated as its own separate simulation, with a new pulling force applied to the molecule. The amount of force used on each window is recorded as a curve reflective of the amount of force needed to build up in order to overcome the restoring within the system. The collective forces recorded from each window is indicative of the energy landscape associated with that system known as the potential mean force (PMF).
For systems consisting of a singular molecule being pulled through a protein, such as a pore or channel, it is possible to calculate its permeability by using equation (2), where ΔG(z) would be the PMF and D(z) could be replaced with known diffusion coefficients of that molecule in water
[22]Park, J., Chun, S., Bobik, T. A., Houk, K. N., & Yeates, T. O. (2017). Molecular Dynamics Simulations of Selective Metabolite Transport across the Propanediol Bacterial Microcompartment Shell. The Journal of Physical Chemistry B, 121(34), 8149–8154. https://doi.org/10.1021/acs.jpcb.7b07232
. One problem with the above method of calculation is that diffusion coefficients of certain molecules in water may not exist within literature. However it is also possible to apply the inhomogeneous solubility-diffusion model to the system, essentially treating it as if it were embeded in a membrane [23]Molecular Dynamics Simulations of Membrane Permeability—PMC. (n.d.). Retrieved from https://www.ncbi.nlm.nih.gov/labs/pmc/articles/PMC6506413/#S2
. As such its is possible to calculate D(z) using Eqautions (2,3) through methods found within literature [24]Ferreira, R. J., & Kasson, P. M. (2019). Antibiotic Uptake Across Gram-Negative Outer Membranes: Better Predictions Towards Better Antibiotics. ACS Infectious Diseases, 5(12), 2096–2104. https://doi.org/10.1021/acsinfecdis.9b00201
[25]Hummer, G. (2005). Position-dependent diffusion coefficients and free energies from Bayesian analysis of equilibrium and replica molecular dynamics simulations. 7, 34–34. https://doi.org/10.1088/1367-2630/7/1/034
[26]Flyvbjerg, H., & Petersen, H. G. (1989). Error estimates on averages of correlated data. The Journal of Chemical Physics, 91(1), 461–466. https://doi.org/10.1063/1.457480
. Unfortunately, we were unable to execute such methods given the large computational time required to setup and perform these methods on our system.
While we are still working towards getting quality MD simulations results, we were able to succesfully run intial simulation through umbrella sampling.
Fig 6. Intial Pull MD simulation used to generate the configurations or windows for Umbrella Sampling
Fig 7. Histograms of sampling from each window. Resveratrol was pull from 2.5 to -2.5 along the Z axis. A total of 25 windows were used with
Additional Projects
Developing ProtKit
We developed ProtKit, an open-source command-line tool for working with Protein Data Bank (.pdb) files. Written in C++17, ProtKit can be run as an executable file on any Windows machine. ProtKit contains a suite of commands for characterizing, manipulating, and combining .pdb files. We are planning future development of ProtKit beyond v1.0 to add support and utilities for files involved in molecular dynamics simulations, like .gro files.
In the development of Manifold, we used ProtKit to visualize protein interactions and make informative graphics. Working on the project gave us a deeper understanding of how protein structure is encoded and how to navigate protein databases like the RCSB Protein Data Bank and UniProt. We hope that future iGEM teams will get value out of and add to ProtKit so that it can remain a helpful structural biology tool to iGEM teams.
Fig 8. Left: A tesselated pattern of PduA homohexamers generated by ProtKit from a .pdb file encoding a single PduA monomer. Imaged with NGL Viewer. Right: The ProtKit script used to generate the .pdb file shown to the left.
ProtKit can be downloaded from GitHub here. More information about ProtKit, including a tutorial, can be found on its website here. Full documentation of ProtKit commands can be found here. The source code can be found in its GitHub repository.
PDE Modeling
We developed a PDE model with a transient metric tensor in order to describe diffusion of proteins in our chassis as it grows with respect to time. Using Finite Element Method in Wolfram Mathematica, we avoided remodeling a tetrahedral mesh at each step of cell growth by implementing a metric tensor whose x-direction evolves with time.
With further development of Manifold, this PDE model can be assessed with respect to actual intracellular dynamics. Additionally, it will be useful to generalize Manifold beyond resveratrol production by simulating alternate metabolite activity.
Fig 9. Finite Element Method tetrahedral mesh for E. coli cell. As the cell grows along its major axis, a transient metric tensor contravariantly "contracts" space so that the cell "grows" without having to redo the mesh in the model.
ProtKit can be downloaded from GitHub here. More information about ProtKit, including a tutorial, can be found on its website here. Full documentation of ProtKit commands can be found here. The source code can be found in its GitHub repository.
Identifying Unique Primers Sequences
Confirming the sequence of synthetic plasmids and parts is a chronically difficult task that must be performed at each step in order to ensure proper assembly of DNA in any chassis. The best, but most expensive, method for determining DNA composition is DNA sequencing. A large amount of the cost associated with DNA sequencing arises in the need to determine primers for Sanger sequencing in DNA to be sequenced; this often requires redesigning parts to terminate in primer binding sites. Additionally, these primer binding sites are only effective if they are unique within the genome of the host cell. If our selected primer binding sequences were to appear in the natural E. coli genome, then Sanger sequencing could become misdirected and fail to actually analyze our DNA of interest.
To solve this issue, we wrote a Python script to predict probabilities of primer conflicts with the entire genome of our chassis. This model saved our team approximately $1000 by allowing us to avoid potential in issues in preparing primer binding sites and the sequencing process.
Fig 10. Graph showing exponential decay of accidental primer similarity probability.
Using NEB's GetSet NEBeta Tool
When making our synthetic assemblies, we extensively used New England BioLabs' GetSet NEBeta Tool tool in order to analyze the quality and reliability of desired ligation results. This use of modeling software informed our experimental design by helping us to choose maximally function ligation sites in our part design.
Using the Salis Lab's RBS Calculator
We used the RBS Calculator from the Salis Lab at Penn State University to model the in vivo performance of ribosomal binding sites (RBS's) in the mRNA transcripts of our synthetic parts
[27]Reis, A.C. & Salis, H.M. (2020). An automated model test system for systematic development and improvement of gene expression models. ACS Synthetic Biology, 9(11), 3145-3156.
. Good RBS efficiency is required for sufficient translation of what plasmids code for into actual proteins.
Creating Informative Models in Mathematica
We created 3D models of the Manifold BMC with Wolfram Mathematica's animation features. This allowed us to better explain what Manifold actually looks like. Additionally, it helped us to visualize the relative sizes of the proteins involved in Manifold, bringing to question a potential concern with the large size of resveratrol pathway enzymes relative to PduA wall proteins.
Fig 11. Mathematica model of Manifold nanofactory. BMC in grey and black, DNA scaffolds in blue, and enzymes in red, green, and purple. The icosahedral structure is a proposed structure for Pdu BMC.
Home
Project
Parts
Outreach
Team
Awards
Resources
Chemistry Building
485 McCormick Rd
CHARLOTTESVILLE, VA 22903
485 McCormick Rd
CHARLOTTESVILLE, VA 22903
