Research Interests, Skills and Projects

In my research, I look at how immune system (IS) response rates and times are influenced by host body size. There are two ways in which body size can influence IS rates and times. a) The IS conducts a very difficult search: it uses very rare antigen-specific IS cells to look for small amounts of antigen in three-dimensional Euclidean space. The search space is much larger in an elephant than a mouse. Hence the IS in an elephant should take much longer to detect the same amount of antigen than in mice. However contrary to our expectations, our empirical results show that the time to detect antigen is nearly invariant. b) The second way in which body size affects IS response is due to the fractal and self-similar branching nature of the cardiovascular system which supplies energy to all cells (including IS cells). Due to physical limitations of these transportation networks, individual cellular metabolic rate is reduced in larger organisms: Bcell ~ M^(-1/4) where M is host body mass. This could imply reduced IS cell proliferation rates and search speeds in larger organisms.

Using an ODE model we showed that IS response rates and times are independent of host body size. We ask what mechanisms has the IS evolved and what architecture does it have to permit such a scale-invariant response. We also investigate whether such strategies can be mimicked to solve search problems in an Artificial Immune Systems (AIS) domain.

Terms and Definitions

Completely Modular System

A completely modular system is a system composed of self-contained autonomous units that do not need to communicate between themselves. The self-contained autonomous units are iterated proportional to the size of the system.

The IS is organized in modular units called protectons which in itself can confer protection. The protecton consists of ~ 10^7 IS cells each of a different specificity and this unit is capable of recognizing the entire universe of pathogens. The protecton as a modular unit is iterated proportional to the size of the organism i.e. an elephant will more protectons than a mice, but the individual protecton is composed of the same number of IS cells.

Completely Modular Detection Network

A completely modular detection network is a network composed of self-contained autonomous units that do not need to communicate between themselves. The self-contained autonomous units are iterated proportional to the size of the system.

We logically extended the concept of modular protectons to propose that the lymphatic network forms a completely modular detection network with lymph nodes (LN) as detection units. We show using an ABM that this IS architecture leads to scale-invariant detection.

Non-Modular Detection Network

A completely non-modular detection network is a network composed of self-contained autonomous units that do not need to communicate between themselves. There are a fixed number of self-contained autonomous units irrespective of system size and they are not iterated proportional to the size of the system.

Our ABM shows that this IS architecture leads to antigen detection times that scale prohibitively.

Hybrid/Sub-Modular Detection Network

A hybrid/sub-modular detection network is a network composed of units that communicate between themselves. The units are iterated sub-linearly to the size of the system.

Our ABM shows that this IS architecture leads to antigen detection times that are nearly invariant and consistent with empirical observations.

Empirical data strongly supports the hypothesis that the lymphatic system forms a hybrid/sub-modular detection network. This raises the question: why is the IS architecture sub-modular when everything else (protecton) is modular? Further analysis shows that the hybrid/sub-modular detection network represents the best solution to the tradeoff between local communication within a module required for rapid antigen detection and global communication to recruit more IS cells and generate a systemic antibody response. If the IS had a completely modular detection network, it would have to communicate less in larger organisms. Such a phenomenon where larger systems communicate less with each other has been observed in microprocessor chips (Rent's Rule).

Extended Protecton

We claim that an "extended protecton" composed of lymph node and its draining region, iterated sub-modularly with organism size is sufficient to function as a unit of protection. It minimizes times to local antigen detection and maximizes the global antibody response.

Some other terminologies that we have introduced are:-

Decentralized Detection Network

A decentralized detection network is a system composed of self-contained autonomous units that do not need to communicate between themselves. The self-contained autonomous units are distributed evenly throughout the system. This is similar to a completely modular system.

Semi-Modular Search

A semi-modular search is the search process that is conducted on a hybrid/sub-modular detection network.

Completely-Modular Search

A completely-modular search is the search process that is conducted on a completely modular detection network.

Keywords: Immune system architecture, extended protecton, protecton, immune system scaling, modular detection network, semi-modular search, immune system communication, lymph node scaling, artificial immune system, immune system modeling, West Nile Virus modeling, WNV modelling, protecton.

Read the paper and another related paper.

The ABM model is here: definition file and init file.

I use a very versatile ABM called CyCells available for free download here.

To run this in CyCells, install CyCells in a folder and paste these 2 files in the same folder.

Then run the following command:

./CyCells -d mys29.def -i mys29.init -t 10000

The -t specifies the number of time steps that it has to run.

The simulation produces a file (test.history) which has the data from the run.

Here is a screenshot of the ABM from our latest paper

Modular RADAR Architecture for Distributed Systems

RADAR (Robust Adaptive Decentralized search Automated Response)

Keywords - scale-invariant detection and response, scale invariant response, scale-invariant detection, immune system scaling, modular search, modular architecture, sub-modular architecture, distributed systems, peer-to-peer systems, artificial immune systems, immune system modelling, intrusion detection systems, malware detection systems, mobile ad-hoc networks, disruption tolerant networks, wireless sensor networks, multi robot control.

Read the paper and another related paper.

Sub-Modular Architecture of Biological Lymph Nodes

Sub-Modular Architecture of Multi-Robot Control Application

Sub-Modular Architecture of Peer-to-Peer System