Week 7 HW: Genetic Circuits Part II: Neuromorphic Circuits

Part 1 HW

Intracellular Artificial Neural Networks (IANNs), also known as neuromorphic circuits, provide significant advantages over traditional genetic circuits that rely on Boolean (digital) logic:

a) Advantages of IANNs

  1. Biological Substrate Compatibility:** While digital logic attempts to force binary “on/off” behavior onto cells, IANNs operate through analog computation, which is much closer to the natural language of biology.
  2. Handling Non-linear Complexity:** Biological systems naturally manage highly non-linear and complex input-output relationships that Boolean logic oversimplifies. IANNs allow for the capture of these non-linearities and non-monotonic behaviors more robustly.
  3. Precision in Decision Boundaries: Unlike digital logic, which only recognizes “high” or “low” thresholds, IANNs can be programmed to respond to specific analog relationships (for example, activating only when two inputs are equal or when a weighted combination exceeds a bias), allowing for much more exact classification of cellular states.
  4. Flexibility and Scalability: The behavior of the circuit can be adjusted simply by modifying the translation rates of the components, allowing decision boundaries to be shifted without needing to redesign the entire system.

b) Useful Application: Cancer Cell Classifier A primary application for IANNs is the creation of high-precision cell classifiers for cancer immunotherapy. Because there is rarely a single “magic” biomarker to distinguish a cancer cell from a healthy one, a sophisticated program is required to evaluate multiple signals simultaneously.

c) Input/Output Behavior

  1. Inputs: The circuit uses multiple intracellular biomarkers as inputs, commonly microRNAs (miRNA). For instance, a classifier can be configured to detect a profile where specific miRNAs (like miRNA-21) are high and others (like miRNA-141) are low.
  2. Processing (Computation): The circuit utilizes elements called “sequestrons” which function as artificial neurons. These perform a weighted summation of inputs: positive signals increase the production of a messenger RNA (mRNA), while negative signals (using proteins like endoribonucleases) sequester that mRNA to prevent its translation.
  3. Non-linear Activation: The circuit implements an activation function known as a ReLU (Rectified Linear Unit). If the negative weighted inputs exceed the positive ones, the output is zero; otherwise, the output grows linearly with the difference.
  4. Output:** If the network classifies the cell as “cancerous” after processing these analog signals, it triggers the expression of a therapeutic agent, such as a killer protein or inflammatory cytokines (e.g., IL-12) to destroy the tumor.

d) Limitations of IANNs Despite their potential, IANNs face several technical challenges:

  1. Node Count Restrictions: There is a physical limit to how many synthetic proteins a cell can express without affecting its viability; most current designs are limited to 10 nodes or fewer.
  2. Environmental Influences and Heterogeneity: External factors such as the immune system and the inherent variability between individual cells can alter circuit reliability and outcomes.
  3. Modeling Complexity: Fully capturing biological realism (such as the spatial clustering of inputs) in computational models remains difficult and often requires multiple iterative cycles of experimentation to achieve accuracy.

Part 2 HW

a) Examples of Fungal Materials and Their Uses Fungal materials, or biomaterials, are grown from microbes (primarily fungi) rather than being industrially manufactured. Some examples include:

  1. Alternative Leathers Used in the fashion industry for “biocouture” and luxury garments.
  2. Luxury Packaging and Insulation:** Due to their ability to grow on agricultural substrates (straw, wood chips), they are used as an alternative to polystyrene (Styrofoam).
  3. Mycelium Bricks: These have been used to build architectural-scale structures, such as the “Hy-Fi” Pavilion at MoMA in New York.
  4. Space Habitats: NASA is investigating their use to build structures on the Moon or Mars, taking advantage of the fact that fungi can be grown in situ.

b) Advantages and Disadvantages Compared to Traditional Materials Advantages:

  1. Superior Insulation:** These materials are thermally and acoustically insulating; blowtorch tests demonstrate that heat does not pass through the material even after a minute of direct exposure.
  2. Lightweight and Sustainable:** They are extremely light and grown from agricultural waste, which significantly reduces the carbon footprint.
  3. Space Logistics:** For space missions, it is far more efficient to transport a small vial of spores than tons of heavy construction materials.

Disadvantages:

  1. Production Time: The growth process is not immediate; it can take weeks for the mycelium to colonize the substrate and months if the fungus is required to fruit to produce spores.
  2. Fragility: While more resilient than some other biomaterials like bio-cement, some states of the material can be brittle before being processed or compressed.
  3. Biological Complexity: Unlike bacteria, the fungi used for materials are genetically distant from traditional laboratory models, making their manipulation more difficult.

c) Genetically Modifiable Functions and Their Purpose Synthetic biology seeks to imbue these materials with new properties that they do not naturally possess:

  1. Physical Properties: Modifying fungi to make the material stronger, more flexible, or more resistant.
  2. Biosensing: Altering the fungus to change color (for example, by producing melanin) in response to pollutants or sugars in the water.
  3. Growth Control: Programming fungi to grow in specific shapes or predetermined patterns.
  4. Extreme Adaptation: For space applications, researchers aim to make them radiation-resistant and capable of feeding on human waste or lunar soil (regolith).

d) Advantages of Synthetic Biology in Fungi vs. Bacteria

  1. Structural Capacity: While bacteria are typically unicellular, fungi form complex networks (mycelium) that act as a biological glue to create solid objects and large structures, something bacteria cannot achieve on their own in the same way.
  2. Neuromorphic Computation: The sources highlight that, unlike digital logic (typical in bacteria), one can implement Intracellular Artificial Neural Networks (IANNs). These allow for analog signal processing that is much more complex and closer to the natural language of biology.
  3. Decision Complexity: With just 10 nodes of an intracellular neural network, a fungal system could perform extremely sophisticated biomonitoring classifications that surpass the simple Boolean logic systems commonly used in bacteria.

Diagram

flowchart LR

%% Inputs
X1["X1"]
X2["X2"]

%% Procesos externos
TX1["Tx"]
TX2["Tx"]

X1 --> TX1
X2 --> TX2

%% Célula
subgraph CELL[" "]

TI["Tl (Csy4)"]
TXm["Tx (Reporter mRNA)"]

SIGNAL["mRNA → Tl"]

%% Regulación
TI --| − | TXm
TXm --> SIGNAL

%% Activación positiva
TXm -->| + | SIGNAL

end

%% Salida
OUT["Fluorescent Protein"]

SIGNAL --> OUT