Week 7: Genetic Circuits Part II

Part 1: Intracellular Artificial Neural Networks

Q1. What advantages do IANNs have over traditional genetic circuits, whose input/output behaviors are Boolean functions?

Traditional genetic circuits are limited to discrete on/off outputs — a gene is either expressed or it isn’t. IANNs can compute analog, weighted combinations of inputs, which means the output can vary continuously depending on the relative strength of multiple signals. This allows a single cell to integrate many inputs simultaneously and produce graded responses, rather than being locked into binary logic. It also means that the same circuit architecture can be tuned to respond to different input combinations just by changing the weights, without redesigning the circuit from scratch.

Q2. Describe a useful application for an IANN; include a detailed description of input/output behavior, as well as any limitations an IANN might face to achieve your goal.

An IANN could be used for early cancer detection inside a cell — taking in multiple biomarker inputs (like elevated levels of specific transcription factors or oncoproteins) and only triggering a reporter output when the weighted combination exceeds a threshold. The input would be endogenous molecular concentrations, and the output would be something like a fluorescent signal or apoptosis trigger. The limitation is that biological noise inside a cell is high, so the weights are hard to set precisely, and the system might fire incorrectly if any one input fluctuates. Translating a trained weight matrix into actual RNA/protein concentrations is also not straightforward.

Q3. Multilayer perceptron diagram

Multilayer Perceptron Diagram Multilayer Perceptron Diagram

The diagram shows an intracellular two-layer perceptron. In this circuit:

  • X1 = DNA encoding Csy4 (endoribonuclease, Layer 1 inhibitor)
  • X2 = DNA encoding Csy4_rec_CasE (CasE mRNA with Csy4 recognition hairpins, Layer 1 output)
  • B1 = DNA encoding CasE_rec_mNeonGreen (mNeonGreen mRNA with CasE recognition hairpins, Layer 2 output)
  • Blue circle = mNeonGreen fluorescent protein (final output)

When X1 is active, Csy4 cleaves the Csy4_rec_CasE mRNA so no CasE is produced, leaving the CasE_rec_mNeonGreen mRNA intact in Layer 2 and allowing mNeonGreen to be translated — the cell glows only when X1 suppresses the intermediate layer.

Part 2: Fungal Materials

Q1. What are some examples of existing fungal materials and what are they used for? What are their advantages and disadvantages over traditional counterparts?

Mycelium-based composites are probably the most well-known example — companies like Ecovative grow fungal networks on agricultural waste to make packaging foam and leather alternatives like Bolt Threads’ Mylo. They’re biodegradable and low carbon footprint compared to plastic foam or animal leather, but mechanical consistency is hard to control at scale and they’re still slower to produce than conventional manufacturing.

Chitosan, most commonly extracted from crustacean shells, can also be derived from fungal cell walls — and the fungal version is gaining interest as a vegan, seafood-free alternative for biodegradable films, wound dressings, and antimicrobial textile coatings. It’s biocompatible and compostable compared to synthetic polymer coatings, but fungal extraction yields are currently much lower than crustacean sources, and the material can be brittle without plasticizers.

Fungal pigments are a lesser-known but really interesting case — species like Monascus produce deep reds and Isaria produces yellows that are being explored as natural textile dyes. Compared to synthetic dyes, they’re biodegradable and less toxic, but yield is low, color consistency batch-to-batch is tricky, and lightfastness is generally worse than synthetic alternatives.

Q2. What might you want to genetically engineer fungi to do and why? What are the advantages of doing synthetic biology in fungi as opposed to bacteria?

I’d want to engineer fungi to secrete specific structural proteins — like silk or reflectin-like proteins — directly into the mycelium matrix as it grows, so the material self-assembles with tunable optical or mechanical properties built in. This is interesting for wearables because you could grow a material that is already functionalized rather than coating it afterward. The reason to use fungi over bacteria here is that fungi are eukaryotes, so they can handle larger, more complex proteins with proper folding and post-translational modifications that bacteria often struggle with. They also naturally grow into 3D fibrous networks, which means you get macroscale structure for free — bacteria just make a liquid culture you then have to process separately.