Week 7 HW: Genetic Circuits Part II

Assignment Part 1: Intracellular Artificial Neural Networks (IANNs)

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

Boolean circuits force biological signals into binary (high/low), but real biomarkers exist at continuous concentrations. IANNs operate on analog values, performing weighted summation and nonlinear activation (ReLU), so they can compute complex continuous functions like bandpass filters and diagonal decision boundaries. These are the kinds of input-output shapes actually needed for problems like cancer classification, where you care about relative concentration levels, not just on/off.

IANNs are also continuously tunable. You can shift the decision boundary by adjusting translation rates, rather than being locked into fixed thresholds. And because they’re universal function approximators, a small number of neurons (even ~10) can approximate essentially any biologically relevant function.

2. Describe a useful application for an IANN; include a detailed description of input/output behavior, as well as any limitations.

Application: targeted cancer therapy using miRNA classification.

Cancer cells have distinct miRNA profiles compared to healthy cells, but no single miRNA is unique to cancer. An IANN could take 2-3 miRNA concentrations as analog inputs, with neurons computing weighted sums where some miRNAs have positive weights (via promoter-driven expression) and others have negative weights (via endoribonuclease-mediated mRNA degradation). By composing a few neurons into a network, you can create a bandpass function that only activates a therapeutic output (e.g. a cytokine like IL-12) when the miRNA profile matches the cancer signature. Healthy cells with different profiles produce zero output.

Limitations:

  • Current circuits support ~3 neurons max, limiting classifier complexity
  • Biological noise blurs decision boundaries, risking false positives
  • Systemic delivery of circuit DNA to cells remains challenging
  • Weights are fixed at design time and can’t adapt to evolving tumour mutations

3. Multilayer perceptron diagram

See hand-drawn diagram.

In the single-layer perceptron, X1 encodes Csy4 endoribonuclease and X2 encodes a fluorescent protein with a Csy4 target site on its mRNA. Csy4 sequesters the mRNA (subtraction), and biology’s floor at zero gives ReLU for free.

For the multilayer version: Layer 1’s output is an endoribonuclease instead of a fluorescent protein. This feeds into Layer 2, where it targets the mRNA of the final fluorescent output. Two chained subtraction + ReLU operations, linked by using an endoribonuclease as the intermediate signal.

Assignment Part 2: Fungal Materials

1. Examples of existing fungal materials, uses, advantages and disadvantages?

Examples:

  • Mycelium packaging/insulation — mycelium grown on agricultural waste (wood chips, hay), packed into molds. Replaces styrofoam. Extremely lightweight and thermally/acoustically insulating.
  • Mycelium leather — processed mycelium sheets as leather alternatives in fashion and biocouture.
  • Mycelium construction — bricks grown from mycelium for architectural structures. The High Five Pavilion at MoMA is built entirely from mycelium bricks. NASA’s Mycotexture lab is exploring mycelium habitats for the Moon/Mars.
  • Biosement — bacteria that convert ammonia to calcium carbonate, solidifying sand/gravel into cement. Companies like Biomason do this at industrial scale.
  • Bacterial cellulose — SCOBY-grown sheets used as fabric alternatives.

Advantages: renewable (grown from agricultural waste), biodegradable, excellent insulation, lightweight, low energy to produce (room temperature growth), mouldable into custom shapes.

Disadvantages: slow growth (days to weeks), weaker than conventional materials, contamination risk during growth, hard to achieve consistent quality at scale, shrinks during dehydration.

2. What might you want to genetically engineer fungi to do? Advantages of synbio in fungi vs bacteria?

Engineering goals:

  • Better material properties (stronger, more flexible cell walls)
  • Biosensing (e.g. colour change in response to environmental signals, like the Aspergillus niger melanin/xylose system)
  • Stress resistance for extreme environments (radiation, low carbon — relevant to NASA’s space habitat work)
  • Programmable growth patterns without external molds

Why fungi over bacteria:

  • Fungi are eukaryotes, so they handle complex protein folding and post-translational modifications better than bacteria
  • Fungi naturally form robust materials (mycelium networks, fruiting bodies) that bacteria can’t
  • They grow on cheap unprocessed substrates (wood, agricultural waste)
  • Food-safe and familiar to consumers, lowering regulatory barriers

Key challenge: most genetic engineering tools exist for model fungi (yeast, Aspergillus) in Ascomycota. The material-relevant species (oyster mushroom, reishi) are in Basidiomycota, a distant phylum. Tools don’t transfer between them, and basidiomycetes are harder to transform because their spores take months to produce. New methods like Agrobacterium-mediated transformation are being developed to bridge this gap.