The Biocomputers: From Biomolecular Circuits To Whole Organism Simulation

INTRODUCTION: 

It's interesting to think about collecting and analyzing data from different sections of our bodies, even individual cells, and using that data to manipulate disease-related changes and regulate biological processes in real time. However, since we are not designed to handle tiny silicon-based devices in our bloodstream, allowing a variety of hardware devices, such as sensors and micro- or nano-computers, to enter our bodies may feel very uncomfortable. However, something external introduction is still required. This "something" should be compatible with our physiology; that is, it should consist of artificially created molecular and cellular systems.

A group of computer scientists and electrical engineers came up with the theory that molecules could compute by comparing their knowledge of theoretical computer science and computer engineering with their observations of how information is handled in living things and cells.

A more serious objection is that the entire idea of "computation" was developed as a formalization of how people think and process information—that is, how the brain functions. Although the brain is capable of computing, it does so through the use of chemicals and cells.

An affirmative answer to these questions was given with the discovery of biological regulation by Jacob and Monod. It turned out that molecules connected into regulatory networks can convert regulatory molecular inputs into specific molecular responses (outputs), and they do so using certain input-output relationships or functions. For example the presence of lactose and the lack of glucose in the bacterium growth medium will induce the Lac Operon and elevate the expression of genes LacA, LacY and LacZ. Roughly speaking, the network computes a logic function (Lactose AND NOT Glucose → LacA, LacY, LacZ). While the real relation is more complicated, the bottom line is clear: the network “computes” a function that connects the concentrations of lactose and glucose in the medium to the Lac operon activity. Therefore what we observe here can be called a biomolecular computation.

Fig. 1: Lac operon as biocomputer

The definition from the Oxford English Dictionary states that a computer is “an electronic device (or system of devices) which is used to store, manipulate, and communicate information, perform complex calculations, or control or regulate other devices or machines, and is capable of receiving information (data) and of processing it in accordance with variable procedural instructions (programs or software). – It simply means computers can do multiple things at a time. This does not mean that all devices can multitask – the world is full of task-dedicated machines, such as cars or CD players.

A Lac network in bacteria can only do one thing very well. Evolutionary pressure and random mutagenesis were shown to alter the Lac network computation. However, in an individual cell the Lac network is not being reprogrammed in a way our personal computers are reprogrammed every time we use a web browser or a text editor. The question arises if and when to call a molecular interaction network a biomolecular computer.

A more far-fetched analogy from a living world is a bacterium cell, with its receptors representing the input peripherals, its entire internal network representing the computer and the cell phenotypic state at a given moment representing the output but it can’t be considered as computer rather a computation as it can’t be altered.

Hence, the biological computer has parts including,

1. Presence of the molecular hardware - the invariant part of the network that can support versatile tasks.

2. Ability to convert well-defined tasks into molecular “machine language” in an automated way.

3. Availability of input peripherals that deliver molecular data from diverse sources and convert them into the molecular “machine format” for algorithmic processing.

4. Availability of output peripherals that convert data from the machine format into diverse biological responses.

5. Availability of a special machine format for data representation.

Fig. 2: Components of Biocomputer

A. In Vitro Systems

I. DNA-based In Vitro Biocomputers:

While DNA is a biological molecule, in nature it normally serves to store genetic information and less as an active participant of reaction networks.

It works in the following ways-

1. Tiling systems

It is a fascinating concept of nanotechnology, where DNA molecules are used like tiny tiles to perform computation and self-assemble into patterns. It was made by Eric Winfree and colleagues.

Fig. 3: DNAs as tiles.

2. State machine systems

Instead of wires and circuits, it uses DNA strands to store states and chemical reactions to change them. Here a DNA is used as one state, another is used as input signal. When both the strands bind, it triggers a reaction like: strand displacement, cutting (enzymes), binding or unbinding.

3. Logic networks

These are just decision making system circuits. These circuits are made of DNAs instead of electronic circuits. An example of it can be an AND gate that uses DNA as circuit.

Fig. 4: DNA based logic gates (AND & OR gates)

II. Protein-based In vitro Biocomputers:

Peptides were proposed as building block for logic gates.

On a chemical-network level, the AND gate was implemented by using two different peptide templates catalyzing the same condensation. The NOR gate was implemented by inhibiting an autocatalytic condensation process independently by two other peptide inputs. Protein-based networks were also implemented in cell-free extracts, where biological processes of transcriptional regulations were reconstituted. This direction is promising for “lab-on-the-chip” applications as well as a way to quantitatively test circuits whose ultimate goal is to operate in cells.

Fig. 5: Protein based logic gate

Enzyme-based circuits have an advantage of being inherently biocompatible. However, it has yet to be shown that these circuits can enable complex programming characteristic of DNA bio-computers. One promising application of these circuits is their utilization as components of larger networks under appropriate circumstances. In addition, developments in enzyme engineering could enable enzyme manufacturing with pre-designed function.

B. In vivo systems

In-vivo systems adapted existing mechanisms for biological regulation, particularly, transcriptional and post- transcriptional regulatory links, and generally adhered to logic circuits as the guiding model of computation. These are simply considered as natural systems that regulate biology. Most of the biological regulation interactions can be classified as either activating or inhibitory systems. Moreover, most of them are subject to saturation (molecular instructions are dependent on the concentrations of the molecules, for instance, the lowering concentrations of Glucose and high concentration of Lactose will trigger Lac operon where lactose inactivates the repressor gene).

I. Protein based in vivo Biocomputers

1. MAPK/ERK Pathway

A chain of protein kinases processes signals step by step. Here the signals are growth factors or stress signals.

Fig. 6: MAP Kinase signaling pathway as biocomputer

2. Calcium signaling network

Calcium acts as input signal, proteins detect Ca⁺⁺ levels to give different responses accordingly.

Fig. 7: Calcium ions as signals at neuromoscular junctions causing skeletal muscle contraction

3. Synthetic protein toggle switch

Here, two proteins inhibit each other.

Fig. 8: Protein activation as toggle switch signaling

4. Protease based logic circuits

Proteases cut specific proteins targets. Used in engineered cellular decision systems.

5. p53 regulatory network

DNA damage levels work as different input signals and in turn, Repair (low damage), or Apoptosis (high damage) come as output signals.

Fig. 9: Pathways linked to p53 gene regulation

II. RNA based in vivo Biocomputers

The combination of RNA information-storage capacity and the fact that RNA can be synthesized in cells make it an ideal substrate for in vivo biocomputers. Briefly, there are two broad categories: riboswitches and small RNAs in RNA interference (RNAi) pathway.

Riboswitches

Natural riboswitches are normally a part of mRNA transcripts and they form locally stable structures (such as stem-loops) in their “ground” state. The ground state can either enable protein translation from that mRNA, or inhibit it. 

Riboswitches have been extensively studied for in vivo computational networks by Smolke and colleagues in their yeast-based networks involved modification of a reporter gene's mRNA to include a number of riboswitches that responded to small molecule inputs and implemented a number of two-input logic gates. 

Fig. 10: RNAs as riboswitches

RNAi

Only a small portion of an mRNA (i.e. “targets”) is required to establish one inhibitory link between the target RNA and the small RNA. The targets can be introduced in arrays in the gene's 3′-UTR, independently on the coding region, implementing NOT-AND-NOT-AND-NOT… logic operations, and multiple mRNAs with identical coding region can be combined in the same network, implementing an OR-OR-OR… logic.

Fig. 11: RNA interference mechanism as biocomputer

Hybrid systems of Biocomputers

Many systems are purely protein- or RNA-based, using hybrid networks that combine both types of elements is increasingly getting more traction. A combination of protein and RNA regulation to control gene expression and in complex synthetic networks has been shown in a number of reports and often found to be superior to either mechanism.

Fig. 12: DNA-RNA hybrid as biocomputer

C. In silico systems

While traditional biocomputing often refers to utilizing biological molecules like DNA and RNA for computing, the broader, more revolutionary application involves using computational intelligence to mimic, model, and eventually govern living systems. These in-silico biocomputers enable scientists to simulate, analyze, and manipulate complex biological phenomena, transforming biology from an observational science into an engineering discipline.

The ultimate goal of in-silico biocomputers is not merely to observe but to design and control. This is the foundation of synthetic biology. Researchers develop, simulate, and optimize complex genetic circuits, such as toggles or oscillators, in-silico before implementing them in living cells. These engineered, living "processors" can then perform tasks inside the body, such as detecting disease markers and delivering targeted therapeutic remedies.

These systems work as follows:

1. Designing Synthetic Genetic Circuits

In-silico systems act as the "Integrated Development Environment" (IDE) for biology. Before a single strand of DNA is synthesized, researchers use software to model genetic logic gates (AND, OR, NOT). By simulating how these gates interact within a bacterial cell, they can predict if a microbe will successfully perform a task—such as changing color in the presence of a toxin—without the trial-and-error of physical lab work.

The same is further verified by actually inserting the same DNA into a cell in the wet-lab to make it act like a computer such as sensing environment, making decisions, and producing a specific output, which acts as customized logical program.

Fig. 13: Synthetic genetic circuit as a biocomputer

2. Metabolic Pathway Engineering

Scientists use in-silico biocomputers to re-map the metabolic "wiring" of microorganisms like yeast or E. coli. By simulating the flow of chemical reactions within a cell, researchers can identify which genes to knock out or add to turn the cell into a microscopic factory. This process is used to efficiently produce biofuels, biodegradable plastics, and complex pharmaceuticals that are otherwise difficult to harvest from nature.

Fig. 14: Metabolic pathway engineering.

3. Folding and Molecular Simulation of Protein 

Programs like AlphaFold represent a peak in in-silico biocomputing. By using AI to predict the 3D structure of a protein from its amino acid sequence, these systems solve a biological "calculation" that would take decades in a traditional lab. This allows researchers to design entirely new synthetic proteins—biological machines that don't exist in nature—to neutralize viruses or break down environmental pollutants.

Fig. 15: Folding and Molecular Simulation of Protein

4. Molecular Docking: The "Lock and Key" Computational Model

Molecular docking, a cornerstone of computer-aided drug design (CADD), acts as a form of biocomputing by simulating the 3D recognition process between small molecules (ligands) and large biological molecules (receptors). While traditional computers use silicon, molecular docking acts as a specialized, "dry lab" biocomputer that uses the laws of physics and chemistry to calculate energy minimization and find the best fit, much like a lock-and-key system.

Fig 16: Docked pose of a drug inside the active site of a target protein

Core Principle: It simulates molecular recognition, looking for the most stable interaction (minimum binding energy) between a ligand and a receptor.

Key Steps:

Preparation: Retrieving/modelling 3D structures (PDB) and preparing the ligand/protein (assigning charges, hydrogens).

Conformational Search: Algorithms (Genetic Algorithm, Monte Carlo) explore thousands of possible poses.

Scoring Function: Numerical evaluation of poses based on steric clashes, electrostatic forces, hydrogen bonds, and solvation effects.

Applications:

Virtual Screening: Rapidly screening large digital libraries (e.g., ZINC, PubChem) to find hits.

Lead Optimization: Modifying lead compounds to improve affinity.

Blind Docking: Performing docking without prior knowledge of the junction site.

Limitations: Traditional docking assumes a rigid receptor ("lock and key"), neglecting the flexibility of proteins during binding. 

 

5. Molecular Dynamics (MD) Simulation: The "Computational Microscope" 

MD simulation models the physical motions of atoms and molecules over time by numerically solving Newton's equations of motion, providing a time-dependent "film" of molecular behavior. 

Fig. 17: MD Simulation of a protein behavior

Video 1: MD Simulation of a protein-ligand complex

(Source: Youtube)

Core Principle: It treats biological macromolecules (proteins, DNA) as dynamic systems in an aqueous medium to simulate physiological conditions.

Key Features:

Flexibility: Unlike docking, MD allows for protein and ligand flexibility, modeling induced-fit mechanisms.

Time-Dependent: It records the location, mode, and rate of movement of each atom at femtosecond resolution.

Force Fields: Utilizes parameters (AMBER, CHARMM, GROMOS) to describe interatomic forces.

Applications:

Refining Docking Results: A-posteriori optimization of structures and calculating precise interaction energies.

Studying Dynamics: Evaluating protein flexibility, stability, and conformational changes.

Binding Kinetics: Designing drugs based on residence time rather than just affinity.

6. Neuromorphic Computing and Brain Mapping

In-silico systems are used to emulate the physical architecture of the brain of creatures like the fruit fly to create neuromorphic chips. This uses massive computational power to simulate the firing of millions of neurons. These biocomputing models help us understand how biological "hardware" processes information with such low energy consumption, leading to AI that learns and adapts more like a living organism than a standard processor. This simulation helps scientists understand the life aspects including behavioral mechanisms, disease models that helps in drug discovery and many more.

Video 2: Whole organism simulation inside the computer

(Source: Youtube)

7. Digital Twins for Personalized Medicine

One of the most powerful applications is the creation of a "Digital Twin"—a comprehensive in-silico model of a specific patient’s biology. Doctors can run thousands of simulations on this virtual biocomputer to see how a tumor might respond to different chemotherapy combinations. This allows for "virtual clinical trials" that identify the most effective treatment for a patient's unique genetic makeup before they ever receive a dose.

Fig. 18: Digital twin simulation inside computer

Fig. 19: Applications of digital twin inside the computer

SOURCES:

1.      Benenson, Y. (2009) ‘Biocomputers: from test tubes to live cells’, Molecular BioSystems, 5(7), pp. 675–685. doi:10.1039/b902484k.

2.      Ausländer, S., Ausländer, D., Müller, M. et al. (2012) ‘Programmable single-cell mammalian biocomputers’, Nature, 487, pp. 123–127. doi:10.1038/nature11149.

3.      Barrett, K.E., Barman, S.M., Brooks, H.L. and Yuan, J.X.-J. (2025) Ganong’s review of medical physiology. 27th edn. New York: McGraw Hill.

4.      Chen, S. and Zaikin, A. (2020) ‘Complex and surprising dynamics in gene regulatory networks’, in Quantitative Physiology. Singapore: Springer. doi:10.1007/978-981-33-4033-6_12.

5.      Cellina, M., Cè, M., Alì, M., Irmici, G., Ibba, S., Caloro, E., Fazzini, D., Oliva, G. and Papa, S. (2023) ‘Digital twins: the new frontier for personalized medicine?’, Applied Sciences, 13(13), p. 7940. doi:10.3390/app13137940.

6.      Cróquer, Z.D.G. and Bermúdez, A.R. (2012) ‘Development of vaccines and gene therapy against HPV infection and cervical cancer’, [Publisher not specified]. doi:10.5772/30034.

7.      Eon Systems (2026) Digital, physics-based fruit fly simulation using full brain emulation. Available at: https://www.youtube.com/watch?v=cvL5NlpYauk (Accessed: 15 April 2026).

8.      GeeksforGeeks (no date) ‘Lac operon’. Available at: https://www.geeksforgeeks.org/biology/lac-operon/ (Accessed: 15 April 2026).

9.      Gewies, A. (2003) ‘Introduction to apoptosis’, ApoReview, pp. 1–26.

10.  Lee, H.-M., Vo, P.N.L. and Na, D. (2018) ‘Advancement of metabolic engineering assisted by synthetic biology’, Catalysts, 8(12), p. 619. doi:10.3390/catal8120619.

11.  Meister, M., Tomasovic, A., Banning, A. and Tikkanen, R. (2013) ‘Mitogen-activated protein (MAP) kinase scaffolding proteins: a recount’, International Journal of Molecular Sciences, 14, pp. 4854–4884. doi:10.3390/ijms14034854.

12.  Methods in Neurosciences (1994) ‘RNA hybridization’.

13.  Nyambo, K., Tapfuma, K.I., Adu-Amankwaah, F., Julius, L., Baatjies, L., Niang, I.S., Smith, L., Govender, K.K., Ngxande, M., Watson, D.J., Wiesner, L. and Mavumengwana, V. (2024) ‘Molecular docking, molecular dynamics simulations and binding free energy studies of interactions between Mycobacterium tuberculosis Pks13, PknG and bioactive constituents of extremophilic bacteria’, Scientific Reports, 14, p. 6794. doi:10.1038/s41598-024-6794.

14.  Prochazka, L., Benenson, Y. and Zandstra, P.W. (2017) ‘Synthetic gene circuits and cellular decision-making in human pluripotent stem cells’, Current Opinion in Systems Biology. doi:10.1016/j.coisb.2017.09.003.

15.  Xu, J. (2025) ‘DNA algorithmic self-assembly’, in Biological Computing. Singapore: Springer. doi:10.1007/978-981-96-3870-3_10.

16.  Zhang, Y., Hu, N., Xu, J. and Wang, Z. (2024) ‘[Article title not provided]’, VIEW, 5, 20230062. doi:10.1002/VIW.20230062.

17. AlphaFold Protein Structure Database (no date) AlphaFold Protein Structure Database. Available at: https://alphafold.com/ (Accessed: 15 April 2026).