Policy context and Community Response

Virginia’s recent changes to HAB advisory guidelines—shifting from dual criteria (cyanobacterial cell counts and toxin levels) to toxin-only thresholds—have raised concerns among SMLA members. Green and Colligan pointed to studies suggesting that cell counts may correlate with toxin production and argued that excluding them could lead to under-reporting. They believe this change may have been driven by funding and logistical constraints rather than ecological evidence.

In response, the SMLA has organized a citizen science program in which volunteers collect water samples and assess them microscopically for HAB species. While this method has limitations, it reflects the community’s commitment to monitoring blooms in the absence of rapid state response. Green and Colligan emphasized that official advisories now only occur if state testing confirms toxins, leaving communities uncertain about the risks posed by dense but non-toxic blooms.

Reflections on Remediation

Although SML’s blooms often resolve naturally, Green and Colligan acknowledged that remediation tools could be valuable for other communities facing persistent or toxic blooms. They noted that current methods, such as algaecide application, carry risks—particularly the potential to lyse toxin-containing cells and exacerbate bloom toxicity.

This feedback supported our decision to explore engineered cyanophages as a targeted remediation strategy. While not a solution for every bloom scenario, phage-based approaches may offer precision and ecological compatibility in cases where intervention is necessary. The SMLA’s input helped us think more broadly about how synthetic biology tools might fit into a larger HAB response framework.

Future Directions

Inspired by our conversation with the SMLA, we see future opportunities to integrate remediation strategies with community-driven monitoring efforts. While our current focus is on phage-based mitigation, future projects could explore how synthetic biology tools might be deployed alongside local initiatives to support real-time, localized HAB response.

Policy context and Community Response

Dr. Egerton described how Virginia’s HAB management framework has evolved since 2001, when microcystin toxin thresholds were first introduced. The policy now includes additional toxins such as anatoxin and cylindrospermopsin. However, due to inconsistent standards in earlier decades, historical bloom data remains difficult to interpret.

He explained that VDH currently relies on ELISA assays to detect microcystins. While effective for confirming toxin presence, these tests are typically conducted after blooms have reached detectable biomass levels. As a result, communities may receive advisories only after a bloom has already formed and potentially dispersed, limiting the effectiveness of response efforts.

Dr. Egerton also highlighted the practical difficulty of coordinating sampling for fast-developing blooms. HABs can emerge within days when conditions align and may dissipate just as quickly due to wind or water movement. This ephemeral nature complicates traditional monitoring and underscores the need for flexible, scalable tools that can support timely decision-making.

Reflections on Remediation

In discussing treatment options, Dr. Egerton noted that remediation remains a controversial and underdeveloped area. Algaecide applications carry ecological risks, particularly the potential for cell lysis to release additional toxins. Biocontrol strategies, such as cyanophage deployment, face both ecological and regulatory hurdles. He suggested that phage-based methods might be more acceptable if used preventatively—before blooms reach high biomass—but emphasized the need for strong evidence of specificity and safety.

This conversation reinforced the importance of developing remediation strategies that are both targeted and ecologically sound. While not universally applicable, synthetic biology tools like engineered cyanophages may offer promising avenues for intervention in cases where blooms are persistent or pose significant health risks.

Future Directions

Dr. Egerton’s insights helped us think more critically about how synthetic biology-based remediation could fit into existing public health frameworks. Future work may explore how such tools could be deployed in coordination with state monitoring programs or community-led initiatives, contributing to a more responsive and adaptive HAB management strategy.

• Current method: Thicker steel plates in high-corrosion zones act as a sacrificial layer
• Drawbacks: Increased overall boat weight; method is visually unappealing
• Implication for our project: Need for solutions that balance durability, cost, and aesthetics, potentially aligning with bio-based or protective coatings explored in soil and marine systems

Ecological Context and Monitoring Practices

Dr. Reece noted that common bloom-forming genera in Virginia include Microcystis, Cylindrospermopsis, and Anabaena. She attributed the rise in HAB frequency primarily to increased monitoring, with warming temperatures also contributing to shifts in species composition and geographic range. Cyanobacteria once limited to southern regions are now appearing farther north, suggesting broader ecological changes.

While she agreed that cyanobacterial cell counts are not reliable indicators of toxicity for human health, she stressed their importance for environmental monitoring. Dense biomass can lead to hypoxia and fish kills, making cell counts a critical metric for assessing ecosystem health at VIMS.

Dr. Reece also described the unpredictable nature of cyanobacterial blooms. Unlike other algae, cyanobacteria lack consistent life cycles, and bloom initiation is heavily influenced by abiotic factors such as sunlight, nutrient input, and temperature. This variability complicates monitoring and response efforts, especially in dynamic estuarine environments.

Reflections on Remediation

In discussing remediation, Dr. Reece offered important cautionary advice. She recommended leveraging the natural specificity of phages to minimize off-target effects and warned against broadening host ranges beyond ecological limits. Her input reinforced the need for targeted, responsible strategies that balance innovation with ecosystem integrity.

These insights helped shape our thinking around synthetic biology-based remediation. Rather than pursuing phage engineering alone, we are considering complementary approaches that align with ecological principles and regulatory feasibility. Dr. Reece’s perspective underscored the importance of designing interventions that are both effective and environmentally sound.

Future Directions

Our conversation with Dr. Reece highlighted the need for remediation strategies that are adaptable to complex and rapidly changing aquatic environments. Moving forward, we aim to explore synthetic biology tools that can be deployed with precision and care, contributing to a broader toolkit for HAB management in Virginia and beyond.

Dr. Wilhelm and Dr. LeCleir prefaced our discussion with some background on Microcystis phage research. The idea that cyanophages might be used to remediate HABs originated in the 1970s, but Microcystis phages have proven difficult to isolate and maintain in the lab.

Importantly, the Microcystis genome comprises about 7% transposable elements, which cause massive genomic variation and likely promote the rapid evolution of phage resistance—even at the level of a single culture or plate. While our team’s preliminary literature review found 13 reports of successful Microcystis phage isolation, the Wilhelm Lab has reached out to each associated research group and found that all but one of the phages (the M. aeruginosa phage Ma-LMM01) were eventually “lost” — i.e., they could not be continually propagated by the scientists that isolated them. The difficulty of propagating Microcystis cyanophages is likely related to Microcystis’s strong anti-phage defense systems, including high rates of genomic rearrangement and prophage immunity provided by viral lysogens. As a result, labs may be “losing” phages as their Microcystis cultures rapidly develop resistance and are no longer useful for phage propagation.

To avoid the host resistance issue, Dr. Wilhelm and Dr. LeCleir recommended that we immediately cryopreserve a stock of any culture that displays evidence of phage lysis in our assays. We could revive this backup stock if our actively-growing cultures became phage-resistant. Were we to find a propagateable phage, they recommended that we prioritize getting it sequenced: Novel sequence data would enhance our currently limited understanding of Microcystis phage diversity (regardless of whether we could consistently propagate the phage itself) and allow us to study the phage’s ecological significance through searches in metagenomic and metratranscriptomic databases.

Given the difficulty of isolating and working with Microcystis cyanophages, Dr. Wilhelm also suggested that algicidal bacteria might represent a promising alternative HAB mitigation tool. Dr. Wilhelm and Dr. LeCleir encouraged us to explore remediation via algicidal bacteria in addition to continuing our search for novel Microcystis cyanophages, which may require a longer-term research timeline.

Selecting a model:

At first, when considering an approach for our predictive model, we considered using a neural network as it is able to handle non-linear data and a large quantity of training data. However Dr. Pasoli brought up a point that with our current training set size, the difference in accuracy would be negligible and other models such as random forest and XGBoost would be better suited (while yielding good results) for our training set.

Preprocessing:

In our discussion, we went over the issues that lie in the metagenomic data publicly available and how that may cause issues in assembling a training set and getting an accurate model. Dr. Pasoli supported the approach of normalizing the values across all samples as a first preprocessing step.

Handling missing inputs:

Another problem that was mentioned was that of unavailable data points that may not have been recorded in a sample. We understand that many models such as a linear regression model require all values to be able to work. Some methods that we found from our literature review showed methods such as removing samples with incomplete data points or inputting zeros in the missing data. We did not want to lose valuable sample data as many of the samples had at least one missing data point value. Dr. Pasoli informed us on his method during his research of taking the mean of the entire column and then filling the missing data point with the mean value. This was helpful in expanding our options and led us to other inputting methods such as k-nearest neighbors (KNN). With XGBoost however we brought up its ability to handle missing values which Dr. Pasoli confirmed was possible. This secured XGBoost as an essential model to test and implement in our project.

Scope of training:

Additionally, we brought up how we had created scripts to read through the Kraken report files and assemble species based on conditions like top n or global n, and we wondered which approach would be more ideal to acquire a broad set of training data that would not lose important species information. Dr. Pasoli noted that, in his experience, he trained his models on all the species found without filtering. This led us to switch gears and instead train our data on every species to ensure that globally-top species and crucial dominating species in a particular sample are not lost.

We went over the process of getting accurate predictors. The presenters emphasized not putting blind faith in AI, despite the many benefits it holds, an understanding of how to verify and ensure correct input data. Statistical approaches like Bayesian statistics, were central to this idea.

Additionally, we questioned how we could implement this into our iGEM project. To which the idea of creating a predictive model was brought up and discussed. The DigiLab team underscored the importance of having good data first both for our predictive model along with any project's direction with AI.

Finally, the meeting went over future implementations, as we bounced ideas informing ourselves on other problems that could have an AI solution– from sequencing data to mathematical modeling. Many aspects of our project and synthetic biology in general had potential, although limited by the availability of data which is needed if a model were to be trained and for patterns to be discovered.

To deepen our understanding of mathematical modeling in synthetic biology within aquatic environments, we consulted Dr. Courtney Harris, a professor at the Virginia Institute of Marine Science. Dr. Harris specializes in 3D modeling of sediment transport and developing numerical models to address environmental challenges. Her expertise in computational modeling of fluid dynamics provided our team with valuable insights into the foundations of modeling complex fluid systems.

Dr. Harris focuses on sediment transport models and advised us that sediment size significantly influences erodibility along the water-earth interface. This concept can be extended to algal mats and other biological materials that form and adhere to the bed of aquatic systems. This concept could be applied to modeling transport of synthetic biological systems that combat erosion.

Additionally, Dr. Harris introduced us to the Shields curve, a graph that outlines the conditions necessary for sediment transport along a riverbed. While our work focuses on bacterial communities rather than sediment, the fluid properties governing sediment transport likely share similarities with those affecting microbial communities at the riverbed. This connection could guide our approach to modeling synthetic biology constructs in these environments.

Dr. Harris' guidance was instrumental in shaping our next steps in creating mathematical models for integrating synthetic biology into water systems. Her expertise in transport dynamics was particularly valuable for developing an understanding of the pressing factors influencing.

One aspect of our project is the prevention and remediation of issues related to flowing water in their natural environments. While we are only doing this in simulated realistic environments in the lab, the future direction of this beyond iGEM would be to introduce these solutions into the real world. Thus, there are policy considerations as well as varying perspectives and stigma surrounding engineered organisms that must be taken into account.

Through VCAP, Emma and Robin mainly interact with homeowners and private land owners facing issues relating to runoff on their land. VCAP currently provides solutions such as increasing plant cover with riparian buffers of native plants to decrease or prevent erosion due to runoff. According to Emma and Robin, they already face some pushback from the community with their current approaches due to misinformation and lack of knowledge which would likely be an even larger issue for attempting to implement engineered solutions. VCAP makes efforts to combat these misconceptions by providing educational resources such as pamphlets and setting up “demonstration areas” where the solution is tested on a small scale to demonstrate how it works.

The discussion extended to the perspectives of professionals in the field, such as Emma and Robin, who would have to interact with and integrate these engineered solutions into the existing prevention and remediation frameworks without being involved in their development. Emma and Robin informed us that they would have no concerns over the introduction of engineered solutions to the field as long as the proper steps for development and approval were followed. They mentioned that homeowners often buy sludge in order to treat their soil, which already contains bacteria. Therefore, if a synthetic biology solution which would require the deployment of bacteria in soil would be offered, they believe that homeowners may be supportive of such a treatment strategy. They also suggested that providing informational materials to professionals unfamiliar with Synthetic Biology would be useful in the implementation of these solutions into existing fields.

Our conversation with Emma and Robin allowed us to understand the perspective of key stakeholders as well as provided us with ideas of educational material that would be useful if we were to actually implement these solutions in real world environments, something that helped us in our development of our Water SynBio guidebook.

• Biofilms often involve multiple bacteria in symbiotic systems
• Standard treatments (chemicals, strong water flow) are largely ineffective
• Charged resins showed some success in breaking down biofilm matrices, suggesting possible avenues for novel interventions