The Automated Recommendation Tool (ART)

Machine Learning (ML) can be quite an intimidating subject for most students unfamiliar with the topic. Our aim is, however, to make our implementation accessible to, and more importantly understandable for, everybody. Therefore, we not only refrained from using complicated ML jargon, we also tried to explain some key concepts behind ML. This way, our Modeling page functions as a crash course in the subject of ML, while simultaneously providing an example of how ML can be incorporated in an iGEM project. We consider both of these aspects to be very useful for future iGEM teams searching for ways of incorporating ML into their own project.

The same strategy was applied when explaining how to use and interpret the code we used for our implementation of the ART. Our Github repository provides easily accessible source code that can be used as inspiration for anybody interested in using the ART. Although some rather difficult subjects are discussed, the language used while discussing it was chosen carefully to be understandable for those with limited experience in statistics, ML, or data science.

Initial background research


Before we settled on our current project idea, in which we focus on the assimilation of ammonia (into a feed additive that reduces ammonia emissions), we first looked into the possibility of enzymatic breakdown of ammonia. More specifically, we were looking into the possibility of creating a biofilter containing microorganisms that could convert ammonia back into atmospheric nitrogen, which is not harmful to nature. For this project idea, we were mainly inspired by iGEM team DTU-Denmark 2013 and iGEM team Virginia 2017. In this section, we explain all the background research we did on this topic, so next iGEM teams can build upon it.

General idea and downsides

In order to break down ammonia (NH3) to atmospheric nitrogen (N2), we would have to engineer the microorganisms to produce several enzymes that would each catalyze a specific step of the step-by-step breakdown. Each enzymatic step would then produce a specific product, which would be used in the next step in the breakdown process. However, we soon found out that this approach had many downsides, making the execution of this project not only infeasible for the timespan we had, but also undesirable for real life application. Hence, we decided to explore other alternatives as a solution to the nitrogen crisis, eventually leading us to our final project idea. The downsides that we encountered were:

  • At least 17 genes need to be cloned to make the pathway work. However, since there are still some uncertainties in which genes are needed for each specific step of the pathway, it is possible that even more genes need to be cloned in addition to these 17. Moreover, in addition to all the genes, you would also need to clone promoters, ribosome-binding sequences and terminators to make the genetic circuit function. Having to clone all these genetic elements in one single organism would not only be challenging, but would probably even be infeasible in the timespan of the iGEM competition. Moreover, one of the essential enzymes, the AMO protein, is a membrane-bound protein, which are notoriously difficult to clone.
  • Some of the intermediate products that would be produced can actually be toxic to the chassis.
  • Some of the intermediate products that would be produced are greenhouse gasses. Therefore, if this project idea were to be executed on a large scale, it would actually contribute to (air) pollution instead of solving it.
  • Due to the characteristics of the enzymes needed for the ammonia breakdown, part of the breakdown would have to happen under aerobic conditions and part of the breakdown would have to happen under anaerobic conditions. This would significantly complicate the final implementation of the process. Moreover, if the enzymes needing to be cloned would be divided over two chassis, each living in a different aerobic environment, a way would have to be found in which the reaction products could be sufficiently transported from one chassis to another and the absorption by the second chassis would also have to be sufficient.
  • Finally, the end product, atmospheric nitrogen, is neutral: it is nor harmful nor beneficial. If we would produce a feed additive, as done for our current approach, we would produce something net positive: something farmers can use to (theoretically) increase the milk production and growth of cows.

Genes that needed to be cloned

In order to enzymatically break down ammonia to atmospheric nitrogen, the following conversions would need to happen:

  • NH3 to NH2OH
    • In order to catalyze this process, the enzyme ammonia monooxygenase (AMO) is needed, which is encoded by amoA1, amoA2, amoB1, amoB2 and petC[1]. AMO is active under both oxic and anoxic conditions. In anaerobic conditions, the enzyme catalyzes the following two reactions: NH3 + N2O4 + 2H+ + 2e- → NH2OH + H2O + 2NO and NH2OH + H2O → HNO2 + 4H+ + 4e- . In the absence of oxygen, the nitrite produced during ammonia oxidation is used as the terminal electron acceptor: HNO2 + 3H+ + 3e- → 0.5N2 + 2H2O[2]. In the presence of oxygen, the AMO enzyme catalyzes the following reactions: NH3 + O2 + 2H+ + 2e- → NH2OH + H2O and NH2OH + H2O → HNO2 + 4H+ + 4e-. This enzyme is a multipass integral membrane protein, occurring in ammonia oxidisers, such as Nitrosomonas. It is a nonamer composed of 3 types of subunits - A, B and C, and contains various metal centers, such as copper, iron and possibly zinc.
  • NH2OH to NO2-
    • In order to catalyze this reaction, the enzyme hydroxylamine oxidoreductase (HAO), which is encoded by 3 genes - hao1, hao2 and hao3[3]. The enzyme has a hemoprotein structure with seven c-type hermes and one specialized P460 type hem per subunit. The oxidation of hydroxylamine (NH2OH) does not depend on oxygen availability and is catalyzed by HAO under both oxic and anoxic conditions. While under oxic conditions, oxygen is used as the terminal electron acceptor; under anoxic conditions, the nitrite produced during ammonia oxidation is used instead. HAO catalyzes the following reaction: NH2OH + O2 → NO2- + H2O + H+. The enzyme is located in the periplasm of nitrifying bacteria[4], such as Pseudomonas.
    • Two cytochromes are also need to catalyze the previously mentioned reactions:
      • Cytochrome c554 - This cytochrome is involved in ammonia oxidation, it accepts electrons directly from hydroxylamine oxidoreductase (HAO). It is encoded by 3 genes - cycA1, cycA2, cycA3 and is located in the periplasm[5].
      • Cytochrome c552 - This cytochrome takes part in the catalysis of the nitrite reduction to ammonia. It is encoded by the cyt gene and is a monomer, which binds 1 heme group and iron as a cofactor. This cytochrome is a probable electron donor to membrane cytochrome oxidase and to periplasmic nitrite reductase. It is located in the periplasm[6].
  • NO2- to NO
    • In order to catalyze this process, the enzyme nitrite reductase (NIR) is needed, which is encoded by the nirS gene. The enzyme catalyzes the following reaction: NO2- + ferrocytochrome c + 2H+ → NO + H2O + ferricytochrome c. In most organisms, nitrite reductase is expressed in the periplasm, but some organisms express it extracellularly or inside membranes[7]. However, in addition to NirS, more unknown genes need to be cloned: at least 10kbp downstream of nirS is needed to have significant NIR activity. If only nirS is cloned, there is only weak NIR[8]. Moreover, the end product of this conversion, NO, inactivates essential cellular enzymes. Therefore, NO is toxic to cells and should be removed from the cell[9].
  • NO to N2O
    • In order to catalyze this process, a nitric-oxide reductase complex (NOR) is needed. To form this complex, 4 genes are needed: 2 genes (norR and rponN) encoding transcription factors and 2 structural genes (norV and norW) encoding proteins/enzymes. norW encodes the cytosolic enzyme NADH: flavorubredoxin reductase, which catalyzes the reaction: oxidized flavorubredoxin + NADH reduced flavorubredoxin + NAD+[9]. norV encodes the cytosolic protein reduced flavorubredoxin, which is consumed in the following redox reaction: reduced flavorubredoxin + 2 NO → oxidized flavorubredoxin + N2O + H2O. The NorV/NorW system therefore consists of a nitric oxide reductase that couples NADH oxidation to NO reduction. The electron transfer chain begins with NADH oxidation by NorW, which then transfers the electron to the rubredoxin domain of NorV [10]. However, for the expression of NOR, which works anaerobically, two transcriptional activators for at least the norVW operon are needed[11]. These transcriptional activators are encoded by the norR gene[12] and the rponN gene[13]. Moreover, the end product of this conversion, N2O is a potent greenhouse gas.
  • N2O to N2
    • In order to catalyze this process, the enzyme nitrous-oxide reductase (N(2)OR) is needed, which is encoded by the nosZ gene. The enzyme catalyzes the following reaction: N2O + 2 ferrocytochrome c + 2H+ → N2 + H2O + 2 ferricytochrome c. In most organisms, nitrous-oxide reductase is expressed in the periplasm, but one organism expresses it in the membrane. In some organisms, the enzyme is functionally expressed under aerobic conditions. However, in other organisms the enzyme gets (completely) inactivated in presence of O2, or activity is 10 times higher if the enzyme is expressed under anaerobic versus aerobic conditions[14].

Background research on the Dutch nitrogen crisis and ammonia production in cattle

For our final project idea, we aimed to tackle the (ammonia part of) the Dutch nitrogen crisis by lowering ammonia production in cattle. To be able to tackle this problem accurately, we did a lot of background research on both the Dutch nitrogen crisis, as well as how ammonia gets produced by cattle and how this production can be lowered. The results of this background research can be found on:

  • Our Project Description page, it describes:
    • How ammonia gets produced by cattle
    • How alpha-amylase should reduce ammonia production in cattle while heightening their milk production and growth
    • Introductory information on what the Dutch nitrogen crisis is and how it harms nature
  • Our Human Practices page, it describes:
    • How much nitrogen (both ammonia and nitrogen oxides) the Netherlands emits
    • The role of agriculture in Dutch and European ammonia emissions
    • International guidelines for maximal ammonia emissions and if the Netherlands is/was achieving them
    • How the Netherlands compares to other European countries on ammonia emissions
    • How ammonia deposition compares to nitrogen oxides deposition
    • The role of Natura 2000 areas in the Dutch nitrogen crisis

Parts collection

Due to the large data set we want to create, we envisioned it is important to think about “How to deal with a large library of constructs”. In order to have a clear overview of the different combinations of promoters, signal sequences, genes and the terminator, a library was developed for 10% of the 640 possible combinations. This library, which can be found on the parts page, was used during the assembly of the constructs so the wet lab team members could keep track of the simultaneous experiments. To make this process easier, each team member was responsible for a certain set of constructs from the parts assembly to the transformation in Saccharomyces spp. In the notebook of the wet lab, this division of constructs and tasks was described. During the assembly and transformation of the cassette plasmids, several wet lab team members worked parallelly, making it possible to assist others when needed and creating a working atmosphere. This way all the experiments ran smoothly and all the cassette plasmids were made within a week.

Since we engineered and assembled a high amount of parts, we registered all the assembled parts on the iGEM parts registry. Secondly, we described all the assembled parts on our parts page. Together with engineering successes, as described on the results page, we can say that most of the cassette plasmids were assembled in the correct way.

Templates to ensure data protection and informed consent

A big part of our own Human Practices work was based on interviews with stakeholders. To ensure that these interviews were handled responsibly and data protection and informed consent were ensured in all our work, we did quite some background research and brainstorming on our data collection, data storage, data protection legislation that concerned us and what would be the best way to handle the ‘aftercare’ of our stakeholders. Since this took quite some time, therefore, preventing us from reaching out to all the stakeholders we wanted to reach out to, we have drafted two flowcharts and a template that can help next years' iGEM teams in ensuring that when they talk to stakeholders that they do so responsibly without having to spend half their iGEM season brainstorming and doing background research. The three documents are meant to be used together: first, you use a flowchart to navigate your questions about your data collection and data storage. Then, you can use a template to draft your information and informed consent sheet. And lastly, you can use a flowchart to guide you through the steps of contacting stakeholders and processing the results. Both flowcharts and the template is discussed in more detail on our Human Practices page, but the documents can also be found here:

Step 3: Flowchart for steps to take when contacting stakeholders

Green initiative


With the same environmentally friendly intention in mind as our project, we wanted to make sure we use up any residual chemicals and kits from previous iGEM Groningen teams in the most sustainable and frugal way possible. Since the University of Groningen has a long history of iGEM teams competing successfully throughout the years many different chemicals, buffers, miniprep and PCR purification kits have been abandoned in storage (Fig.1). This year, after the past iGEM season when no one used the iGEM laboratory, we took upon ourselves to clean up and organize the big storage units filled to the brim with reagents, most of which were still within the expiration date (Fig.2).

Before - The abandoned storage with old and used kits, chemicals and other materials from previous iGEM Groningen teams

Figure 1: Before - The abandoned storage with old and used kits, chemicals and other materials from previous iGEM Groningen teams

After - The now organized storage

Figure 2: After - The now organized storage

Testing of old miniprep kits

In order to examine the functionality of miniprep kits, used for purification of plasmids from our chassis, simultaneous experiments were conducted using a brand new kit and a previously used kit. In order to observe purification differences using the different kits, the DNA concentration was measured using a Nanodrop spectrophotometer, quantifying the DNA concentration at a spectrum of wavelengths.

Table 1: Nanodrop measurements comparing old and new miniprep kits​​




28 - new



28 - old



44 - new



44 - old



46 - new



46 - old



47 - new



47 - old



MilliQ - new



MilliQ - old



The A260/280 ratio represents the purity of the DNA, and ideally, the value would be approximately 1.8. The samples measured after using the old kit have seemingly correct values for this ratio, as can be seen in Table 1. However, the total DNA concentration is surprisingly highcompared to the results obtained when using the new kit. To make sure that the buffers from the old kit were not contaminated with DNA, MilliQ was used as a control. As is evident in the results table, DNA could be measured in the MilliQ samples using both kits. Nevertheless, the concentration of DNA in the MilliQ sample using the old kit was 6-fold higher in comparison to the new kit. Therefore, it can be concluded that the buffers from the old kit most likely were contaminated with DNA. In order to use an outdated miniprep kit in a safe way, the above-mentioned experiment should be conducted where MilliQ is used as a reference sample.


  1. Ammonia monooxygenase in UniProtKB. (n.d.). UniProtKB.
  2. Schmidt, I. (2008). Nitric Oxide: Interaction with the Ammonia Monooxygenase and Regulation of Metabolic Activities in Ammonia Oxidizers. Nitric Oxide, Part F, 121–135.
  3. Hydroxylamine oxidoreductase in UniProtKB. (n.d.). UniProtKB.
  4. Arp, D., Sayavedra-Soto, L., & Hommes, N. (2002). Molecular biology and biochemistry of ammonia oxidation by Nitrosomonas europaea. Archives of Microbiology, 178(4), 250–255.
  5. cycA1 - Cytochrome c-554 precursor - Nitrosomonas europaea (strain ATCC 19718 / CIP 103999 / KCTC 2705 / NBRC 14298) - cycA1 gene & protein. (n.d.). UniProtKB.
  6. nrfA - Cytochrome c-552 precursor - Escherichia coli (strain K12) - nrfA gene & protein. (n.d.). UniProtKB.
  7. “Information on EC - Nitrite Reductase (NO-Forming).” Brenda. (October 8, 2021).
  8. Ohshima, Takayuki et al. 1993. “Cloning and Sequencing of a Gene Encoding Nitrite Reductase from Paracoccus Denitrificans and Expression of the Gene in Escherichia Coli.” Journal of Fermentation and Bioengineering 76(2): 82–88
  9. “Escherichia Coli K-12 Substr. MG1655 NADH:Flavorubredoxin Reductase.” EcoCyc. (October 8, 2021).
  10. “Escherichia Coli K-12 Substr. MG1655 Anaerobic Nitric Oxide Reductase Flavorubredoxin.” EcoCyc. (October 8, 2021).
  11. “NorR - Anaerobic Nitric Oxide Reductase Transcription Regulator NorR - Escherichia Coli (Strain K12) - NorR Gene & Protein.” Uniprot. (October 8, 2021).
  12. “NorR - Anaerobic Nitric Oxide Reductase Transcription Regulator NorR - Escherichia Coli (Strain K12) - NorR Gene & Protein.” Uniprot. (October 8, 2021).
  13. “RpoN - RNA Polymerase Sigma-54 Factor - Escherichia Coli (Strain K12) - RpoN Gene & Protein.” Uniprot. (October 8, 2021).
  14. “Information on EC - Nitrous-Oxide Reductase.” Brenda. (October 8, 2021).