Team Attributions Collaborations Inclusion

Contribution Description Engineering Success Implementation Notebook Proof of Concept Experiments Database Modeling

Human Practices Education and Engagement

Safety

Kill Switch Modeling

Introduction

In addition to creating an online database of available biosafety related parts, we designed a modeling tool for teams to see how effective their chosen sequences may be in their system.

Original Plan

We originally planned to create a model that would predict the escape rate and whether or not the parts we chose were the best for our host. After talking with other teams that have had experience or have been currently attempting to include a biocontainment mechanism into their project, we knew this modeling tool would be helpful to them as well.

We originally planned to find the concentration of the protein that would trigger the cell’s death by using a gene expression equation. We were then going to determine how much of the protein was needed to kill the cell and then simulate a colony by factoring in the escape rate to see how the whole system would react.

However, most of this plan focused on experimental data that we were unable to collect since we did not have lab access due to COVID19.

Revised Plan

When we presented our original plan to Caleb Bashor, he suggested that a more effective model would be to focus on what would render the kill switch ineffective. In particular, he suggested that we determine how likely it is that a mutation would occur in the system. We decided to focus on modeling the individual probability of a single nucleotide mutation coding for the incorrect amino acid, as described in An Analysis of Single Nucleotide Substitution in Genetic Codons - Probabilities and Outcomes.

Assumptions

Each base pair has the same probability of mutating to any of the other base pairs.

Procedure

Code for modeling can be found in our Modeling GitHub Repository. We created a Java program that analyzes a sequence input to determine the individual probability for each nucleotide triplet by counting the number of single codon mutations that would result in the incorrect amino acid and dividing that by the total number of single codon mutations. The program also averages the results for each nucleotide triplet to create an average probability for the sequence.

Results

We used our program to analyze the eight toxins that we used in our plasmid designs. The plasmid names, information, and average probability are shown below for each of the toxins. We will then use this data when designing our plasmids.

Name	Average	Type	Description	Source	URL
T4 holin	0.7872	Toxin	Sucide Mechanism	iGEM	BBa_K112000
hokD	0.7632	Toxin	holes in the membrane	iGEM	BBa_K1497008
ParE	0.7632	Toxin	inhibits DNA replication	iGEM	BBa_K2150104
ghoT	0.8087	Toxin	puts holes into the membrane	iGEM	BBa_K2919000
Barnase	0.7666	Toxin	eats up both RNA and DNA	iGEM	BBa_K1172904
NucD	0.7539	Toxin	holes in the membrane	iGEM	BBa_K3317053
ccdB	0.7721	Toxin	Interrupts DNA gyrase and kills the cell	Lit	https://www.sciencedirect.com/science/article/pii/0378111994902356
mazF	0.7594	Toxin	Cleaves mRNA to stop protein synthesis	Lit	https://www.sciencedirect.com/science/article/pii/S1097276503004027

We also used Excel to visualize each of our sequences by using conditional formatting to show the probabilities for each nucleotide triplet in relation to the other nucleotide triplets. The color ranges from red to yellow to green, where red is a higher probability, yellow is a middle probability, and green is a lower probability. The higher the probability, the greater the chance of mutation. The image below shows each sequence running horizontally from left to right. Each square represents one codon. The eight sequences are show in the same order that is presented in the table above.

Future Work

We would like to add weights into our modeling to account for the likelihood of one nucleotide mutating to another nucleotide, since in our model we assume that all nucleotides have the same probability of mutating into any other nucleotide. We’d also like to add in other types of mutation such as insertion and deletion to further refine the model.

We would also like to gather lab data to continue our original modeling. We would use lab data from multiple different killswitch constructs to determine the relative values for the parameters so that teams can implement our modeling into the design of their project, instead of using it as a launching point for their own modeling.

We would also expand upon the model to create a simulation for each killswitch by using stochastic modeling to depict the effect that each individual kill switch trigger (“on” switch) has on the population as a whole. From there we would have a way to simulate the effectiveness of various constructs. We would also analyze these scenarios to determine which parameters have the greatest effect on the system to determine which components of the kill switch need to be improved.

HP Modeling Steps

GMO Modeling

Vitamin A deficiency is a world-wide problem, especially prevalent in developing countries, that is known to cause symptoms such as stunting, vision loss, muscle loss, and more. The prominence of these deficiencies in developing countries is enhanced when weather conditions are not permissive to regular crop growth, which was the case in the 2019 crop season for Zimbabwe. Due to poor weather conditions, Zimbabwe lost about 40% of their maize, causing them to have less than 50% of the national maize consumption requirement: http://www.fao.org/3/ca6624en/ca6624en.pdf. Moreover, maize is one of the main sources of vitamin A in Zimbabwe, leading to a growing level of vitamin deficiency. To combat the growing levels of vitamin A deficiency from the loss of maize, it is important that products such as GMOs are used to provide the recommended levels of nutrients. One GMO in particular, transgenic multivitamin corn, has the capacity to fight vitamin A deficiency levels if implemented. This corn variety has 169-fold the normal amount of Beta-carotene (Beta-carotene is metabolized by the body to form Retinol, which is a form of vitamin A) or about 60 micrograms of Beta-carotene per one gram of maize (almost double the amount of Beta-carotene found in Golden Rice), so even with deficient maize production, the amount of vitamin A from the Beta-Carotene is substantial for the population: https://www.pnas.org/content/106/19/7762

The first step in producing our GMO analysis models was to breakdown the population based on age and gender (data found with Population Pyramid: https://www.populationpyramid.net/zimbabwe/2019/). Our groups were broken down based on the National Institutes of Health-Office of Dietary Supplements recommended vitamin A daily intake values. We did not separate based on gender until the age of 14 because recommended vitamin A values do not change based on gender in those who are 13 years old or below. Additionally, when we were breaking down the population, we used the population pyramid website to find the percentages of those who were in a certain age range relative to the overall size of the population (example: 0-4 year olds made up 7.3% of the population in 2019). Further, in order to break the age groups up in a way to correctly distribute them based on necessary vitamin A levels, we took fractions of the age range percentages to estimate the value of how many people would need specific amounts of nutrients. For example, when we were accounting for how much of the population is made up of those aged “0-6 months” for 2019, we took the percentage breakdown from the population pyramid (for 0-4 years: 14.6%) and multiplied this by 0.0125 to assume that the age groups made up even segments of the percentages. Another example would be for 1-3 year olds in 2019, which we used the 0-4 year range for (14.6%). Since we would say that 1-3 makes up 50% of the 0-4 year range, we would take 14.6% and multiply it by 0.50 to get our percentage of 1-3 year olds. After this, we took our percentages and multiplied them by the total population.

2018
Age (Years)	Percent of Population	Number of People
0-0.5	1.9%	274,337.4
0.5-1	1.9%	274,337.4
1-3	7.6%	1,097,349.7
4-8	14.825%	2,140,553.9
9-13	13.125%	1,895,094.1
14-i (male)	27.875%	4,024,818.8
14-i (female)	32.775%	4,732,320.6
Total	100%	14,438,812.0

2019
Age (Years)	Percent of Population	Number of People
0-0.5	1.825%	267,279.9
0.5-1	1.825%	267,279.9
1-3	7.3%	1,069,119.5
4-8	14.825%	2,171,191.4
9-13	13.25%	1,940,525.2
14-i (male)	28.0%	4,100,732.4
14-i (female)	32.975%	4,829,344.7
Total	100%	14,645,473.0

Source: https://www.populationpyramid.net/zimbabwe/2019/

Vitamin A recommended daily intake values based on age:

Age (Years)	Vitamin A Recommended Daily Intake (mcg)
0-0.5	400
0.5-1	500
1-3	300
4-8	400
9-13	600
14-i (male)	900
14-i (female)	700

Numbers from: https://ods.od.nih.gov/factsheets/VitaminA-Consumer/

In order to determine the required amount of vitamin A to sustain the population of Zimbabwe, we took our population breakdown and multiplied each age group by the recommended vitamin A levels to ensure the GM maize provides substantial vitamin A.To get the total daily and yearly values in micrograms, we added up the amount of micrograms recommended for each age group (daily) and then multiplied that number by 365 (yearly).

2018
Age (Years)	Percent of Population	Number of People	Vitamin A Recommended (mcg/day)	Vitamin A Intake for Population (mcg/day)
0-0.5	1.9%	274,337.4	400	109,734,971.2
0.5-1	1.9%	274,337.4	500	137,168,714.0
1-3	7.6%	1,097,349.7	300	329,204,913.6
4-8	14.825%	2,140,553.9	400	856,221,551.6
9-13	13.125%	1,895,094.1	600	1,137,056,445.0
14-i (male)	27.875%	4,024,818.8	900	3,622,336,961.0
14-i (female)	32.775%	4,732,320.6	700	3,312,624,443.0
Total	100%	14,438,812.0	3800	9,504,347,999.0

2019
Age (Years)	Percent of Population	Number of People	Vitamin A Recommended (mcg/day)	Vitamin A Intake for Population (mcg/day)
0-0.5	1.825%	267,279.882	400	106,911,952.9
0.5-1	1.825%	267,279.882	500	133,639,941.1
1-3	7.3%	1,069,119.529	300	320,735,858.7
4-8	14.825%	2,171,191.372	400	868,476,548.9
9-13	13.25%	1,940,525.173	600	1,164,315,104.0
14-i (male)	28.0%	4,100,732.440	900	3,690,659,196.0
14-i (female)	32.975%	4,829,344.722	700	3,380,541,305.0
Total	100%	14,645,473	3800	9,665,279,906.0

Vitamin A (mcg) Needed	Daily	Yearly
2018	9,504,347,999	3.47x10^12
2019	9,665,279,906	3.47x10&12

We then decided to analyze the vitamin A produced from non-GMO maize in 2018 and 2019, and we also modeled the hypothetical amount of vitamin A that would have been produced in 2018 and 2019 if the maize was the GMO variety. We used the FDA’s Beta-Carotene to Retinol conversion factors to determine how much vitamin A is in one gram of GMO maize: https://www.fda.gov/media/129863/download (page 17). We then used these values to determine how many micrograms this provides for a person in a year before breaking it down to the daily amount. As we can see, the vitamin A levels with GMO maize are significantly higher than vitamin A levels in non-GMO maize.

Once we determined the amount of vitamin A in micrograms from the difference in maize varieties and in years, we determined the deficiency or surplus of vitamin A by subtracting the necessary value (calculated in step 3) from the actual value of vitamin A (calculated in step 4). We took this value and divided it by 365 to determine the daily deficiency/surplus and then divided it by the population size for the specific year to determine the deficiency/surplus for each person daily.

As we can see from the models, GMO maize has the potential to have significantly positive impacts in Zimbabwe if implemented. Despite this, the complete implementation of GMO maize will not happen overnight, and there may be pushback from those who are opposed to GMOs, with some saying that GMOs do not actually derive a nutritional benefit. Our models help to showcase just one aspect of why GMOs should be more widely accepted and utilized in developing countries to create a sustainable future for all.
It is also a common misconception that there is danger associated with GMO foods, but this is not the case. GMO foods are completely safe and may even be safer than non-GMO foods because of the process they undergo in development. The genes added to these foods can increase nutrient levels and can help pave the way for a sustainable future.

GMO Modeling Collaboration

As part of a collaboration with FCB-UANL, we sent them our GMO modeling and they followed our method to create a similar model for Mexico. You can find a copy of their modeling and results here.

Team:OhioState/Model