Q: How can biological filaments perform computing operations?
A: The filaments explore networks that encode a mathematical problem. The filaments move from the entrance to several exits according to traffic rules determined by the the network(e.g. in subset sum, turning left means adding a number). The information carried by the filaments (e.g. spatial postion, tag information) will provide the solution to the problem. For example, the exit at which filaments arrive corresponds to the total sum of all numbers added to the subset.
Youtube video explaining how biological filaments solve the subset sum problem
Q: What is the difference between biocomputation, biological computing and computational biology?
A: In biocomputation or biological computing, biological components like proteins, DNA or cells are used to perform mathematical operations as well as storing and retrieving data. In computational biology, computer programs are used to model biology.
See Wikipedia page on biological computing
Q: What is the advantage of network-based biocomputation?
A: there are two advantages, (1) the energy efficiency and (2) that all computing operations are performed in parallel.
The energy efficiency is due to the fact that biological computing is driven by ATP. ATP (adenosine triphosphate) is a high energy molecule common to all living cells. In biocomputation, the energy of ATP is used to power the motility proteins that propel the filaments through the maze. The conversion from chemical energy to mechanical energy and the movement of the filaments has been optimized by evolution and is extremely efficient.
Parallel computing: the filaments are exploring all possible paths in the network at the same time and, thus, each computation process is independent and in parallel. In contract, a serial computer would have to test all possible solutions one at a time.
Q: What are the biggest challenges in network-based biocomputation?
A: The biggest challenge is to upscale the networks in order to outcompete other alternative computing technologies. This can be broken down into four components:
1. Algorithms: The first step is to code algorithms into physical networks and optimize them so that they can be explored by filaments. This includes also formal verification, so we can be sure that the mazes work before we waste a lot of resources for the fabrication.
2. Architectural elements: before the networks can be fabricated, the design of the hardware needs to be optimized. This defines the traffic rules in the mazes and makes sure that they are followed by the filaments e.g. angles and curvatures in split junctions are designed so that the filaments stochastically turn left or right. The error rates being low is a precondition for making bigger networks.
3. Scaling the biology: firstly, we need to make sure that there are a sufficient number of agents to explore all network channels and exits. Secondly, for the filaments to carry information about the path they have taken through the network, so-called path-tagging, facilitates upscaling.
4. Fabrication and integration: here, upscaling is not only about size but also reducing error rates and developing new microfabrication methods that can integrate several materials and are at the same time friendly to the biological molecules.
The research challenges in network-based biocomputation
Q: What means tagging?
A: The filaments can be tagged via attaching certain labels or modifications along the path they have taken through the network. This is important in upscaling of biocomputation because networks can be designed in a more compact way. A solution how the filaments can pick up colored tags on their way through the maze at coloring stations has been proposed by Adam Micolich as a contribution to the 1st Bio4Comp Open Innovation Award, an idea competition that was held by Bio4Comp in 2019. Adam Micolich’s contribution “Colorimetric path tagging of ﬁlaments using DNA-based metaﬂuorophores” was the winning idea.
1st Bio4Comp award and winning idea Colorimetric path tagging of ﬁlaments using DNA-based metaﬂuorophores