Bio-IT

At the Interface Between Biology & Information Technology

Henrik I. Christensen

1. Summary

This report summarises the proposal for an action plan in the field of ‘Bio-IT’, which was developed collectively by the participants of the ‘Bio-IT’ workshop, in Brussels, 3-4 April 2000.   

The programme explores two major lines of research termed ‘Hybrid Systems’ and ‘Exploiting Embodiment’. Examples of research issues to be addressed are listed in Annex I.   

The programme draws it inspiration from biology. Knowledge of biology is used for the construction of both artifacts and hybrid systems. The programme is not limited to a particular business sector or a particular key-area but covers aspects from handicap aids and computer interfaces to new paradigms for computing and system design. While the proactive initiative ‘Machines that live and grow’ involves a two-way interaction between life science and computer science, this programme will primarily use methods from life science as a source of inspiration for the construction of systems.

2. List of Participants

Dr Martyn Amos, University of Liverpool, UK; Dr Rita Casadio, CIRB Biocomputing, IT; Prof. Henrik Christensen, KTH, SE; Dr Ron Cottam, VUB, BE; Dr Ulf Dieckmann, IIASA, AT; Prof. Dr Georg Dorffner, University Wien, AT; Prof. Dr Gusz Eiben, Virja Universiteit Amsterdam/Leiden University, NL; Prof. Massimo Grattarola, University of Genova, IT; Dr Inman Harvey, University of Sussex, UK; Dr Henrik Lund, Aarhus University, DK; Dr Alvaro Moreno, ES; Dr Wim L.C. Rutten, University of Twente, NL; Dr Martin Schütz, CASA, DE; Dr Alain Strambi, CNRS, FR. From the EC: Pekka Karp, Christina Versino, Jaques Lacombe, Jose Luis Fernandez Villacanas Martin, Ramon Compano, and Simon Bensasson

3. Background

Today information technology systems are brittle and often non-robust. In contrast, biological systems typically exhibit a high degree of robustness and a much larger degree of adaptation to changes in the environment. Through studies and inspiration from biology, it may be possible to provide new methods for design of IT systems that exhibit improved performance. In addition, it might be possible to build hybrid systems that combine biological and IT systems. The aim of this workshop was to explore different approaches to the design of Bio-IT systems. This report summarises the major ideas identified at the workshop but does not by any means exhaust the range of issues covered.

4. Objectives of ‘Bio-IT’ action plan

The area of Bio-IT brings together the topics of biology, information technology, and systems engineering, as shown below:     

Two areas were identified as the basis for the Bio-IT initiative: ‘Hybrid Systems’ and ‘Exploiting Embodi-ment’. These two areas are outlined below.

At the interface between biology and IT there are a number of interesting new opportunities. In vitro cultured neuro-electronic systems will enhance our understanding of neural information processing and adaptation. Biological systems (e.g. sensors) have a wide range of uses and exhibit a remarkable level of adaptation to the environment. So far, limited success has been achieved in the design of artificial sensors and systems with similar characteristics. Can biological sensors be grown for use in artificial systems? How can computers be integrated into biological systems to augment systems or to replace neural circuitry or even “to train” them towards information processing tasks? Both bio-robotics and bionics are prototypical examples of systems representing this type of research. The expectation is that the hybrid approach will enable new in-vivo and in-vitro systems that exhibit significant advances for living and artificial systems in terms of richness, plasticity and robustness.   

Key-questions to be addressed in relation to hybrid Bio-IT systems include:

• Methods for growing and interfacing biological sensors and systems to artifacts such as robots
• Methods for interfacing artificial sensors to biological nerve tissue to enable augmentation of such systems
• Can biological insight gained from in vitro cultured neural networks on electronic substrates be used for the construction of new types of large-scale artificial neural network?

Another area of interest for hybrid systems is molecular computing. Today molecular computing/DNA computing has been demonstrated at the methodological level, and interesting theoretical results have been reported. Through new research in proteins, basic theory and process technology it can be expected that molecular computing can be made a functional paradigm for massively parallel computing that would enable problems currently considered intractable to be addressed. An interesting problem is also the integration of traditional computers with molecular computers and automatic scheduling of jobs across the two paradigms.

Key questions to be addressed in relation to molecular computing include:

• Design of new proteins that are particularly well suited for computing.
• Interfacing of molecular and traditional computing paradigms
• Operational (limited size) process technology for molecular computing

The second area identified is related to exploitation of embodiment. A problem exhibited by many IT systems is a lack of robustness and adaptation. At the same time it is well known that biological systems exploit these characteristics to enable survival. The problem experienced in current IT systems are in part attributed to lack of a holistic approach to the design of systems and to use of technologies that lack self-adaptation. There is thus a need for studies of how embodiment influences the systems at all levels from individual units to the complete systems. This includes questions related to emergence of functional complexity, strategies for co-design of integrated systems, and self-maintenance and self-reproduction. Many of the issues mentioned for this particular area are closely related to the proactive initiative of ‘Machines that Live and Grow’

Key questions to be addressed in relation to the embodiment of systems include:

• Holistic methods for design of large-scale artifacts (i.e. complex systems)
• The emergence of functional complexity in relation to the body
• Strategies for co-design of integrated systems
• Mechanisms for self-repair and self-(re)production
• Studies of biological systems to understand human constructed systems such as the World Wide Web.

The objectives of this programme are thus three fold:

• Studies of hybrid systems where neuro-biological tissue and computers are integrated to provide bio-artifacts such as robots or for bionics. It is of particular interest here to go beyond integration of simple systems to allow construction of fully operational systems of realistic complexity.
• Studies of the basic mechanisms for molecular computing to enable construction of an operational and deployable paradigm that can be used in the context of large-scale computing.
• Studies of systems in terms of their embodiment to enable understanding of systems that have a complexity comparative to biological entities or communities.

5. Potential innovation

Interfacing the nervous system to artificial systems is already taking place today, but only on a very limited scale. Studies in functional electrical stimulation (FES) are used to help handicapped patients regain mobility, but the results are so far limited to simple motion, and the level of control provided is limited. In terms of building hybrids Japanese researchers have showcased a hybrid cockroach, but the system exhibited little or no deliberate behaviour. Research in this domain would enable exploitation of biological sensor neuronal control systems to gain robustness, increased range of sensing, and it would at the same time provide valuable insight into the operation of sensor-motor information processing. This is a topic that may also be of interest to life-science. Use of artificial sensors for interfacing to neuronal cell populations as well as brain tissues would pave the way for augmented sensing, which could be used for applications such as improved driving and flying performance, or for new computer interfaces (in particular for ubiquitous computing). In addition, the development of new bionics technology might provide valuable technologies to be used for rehabilitation engineering and providing assistance to the handicapped.   

In molecular computing the long-term goal is considered to be a desktop size co-processor interfaced to a regular computer for hybrid computation. The introduction of desktop molecular computing would enable large scale parallel computing that can be used to address problems that today are considered intractable. In turn, it would imply a change of paradigm in computing and have significant impact on diverse areas of computing. An example would be encryption, where the standard 128 bit coding, used today, could be easily deciphered.   

The embodiment of systems will in particular study complex systems. Studies of biological systems will allow insight to be gained to enable the design and understanding of large-scale systems. Today complex systems such as the internet are considered chaotic and beyond comprehension. Through studies of large-scale systems such as ant colonies it may be possible to gain an understanding of group behaviour that can be used for studies of complex artifacts. In addition, studies of self-adaptive and self-productive systems could enable the design of new generations of micro-processor technology that are robust to minor faults, which would mark a new era in reliable computing and robust systems. The study of system embodiment would also enable new methods for system level design through an understanding of the coupling of holistic and local computation, which in turn would allow formulation of tools for improved system design and analysis. While the ‘Machines that Live and Grow’ programme is expected to provide a close interaction between life science and IT, this programme will draw its inspiration from studies of biological systems, although providing feedback to life sciences is not the primary objective, and it will thus be possible to extrapolate beyond biological systems.

Annex I: Example of research areas

In-vitro use of biological sensors/tissue: Biological sensors/tissue (nerve tissue) demonstrate a remarkable plasticity which is far beyond that achieved in artifacts today. It is thus of interest to study how biological sensors/tissue can be grown for use in artifacts. The interfacing of biological tissue to computers will provide a range of new sensory modalities that are richer and more versatile than those available today. An example is in terms of olfaction where specific silicon sensor today can be used for detection of specific substances but the versatility experienced in biological systems cannot be replicated. The same is true in ultra-sonic ranging used in bats, or the richness of sound perception. The studies of in-vitro use of nervous tissue (e.g. networks of cultured neurons) would at the same time provide new insight into biological tissue and sensors, which for example might enable construction of new types of computers or sensor systems. The study would at the same time enable construction of more versatile artifacts such as bio-robots (or even humanoids)

In-vivo use of computer technology (bionics): Interfacing computer to living (neural) tissue is an interesting issue as detailed understanding of the control and communication mechanism in biological systems would allow augmentation or enhancement of such biological systems. The obvious example is construction of systems that allow handicapped to compensate for various impairments. In addition, the interfacing of computer would allow use of new sensory modalities to improve the performance of humans in various situations and it would enable a new generation of human-computer interfaces for a range of different applications.

Proteins for molecular computing: Through careful design of new types of protein specifically engineered for computing it might be possible to speed up the process and miniaturise the overall process to enable use in a desktop environment. This would open up a new paradigm in massively parallel computing. The design of the protein would be linked in advance to the process of molecular computing to its enable use in a non-laboratory environment.

Interfacing molecular computing: Current state of the art allows use of molecular computing in a laboratory setting. For traditional applications there is a need for automatic interfacing of such processes to traditional computer technology or alternatively there is a need for new programming methods to enable data entry, problem specification and data interpretation to allow regular users to exploit the advances in molecular computing.

Strategies for co-design of integrated systems: Most systems engineered today do not exploit a holistic method for design, construction, deployment and use. By studying biological counterparts, it can be expected that embodiment (sensing, actuation and morphology) and computation can be integrated in a more efficient manner. The studies of biological entities and communities will at the same time allow studies of complex systems, such as the internet, as a living entity. The study of the relationship between components and systems will also provide new insight into systems engineering.

Methods for self-repair and self-production: Current technology is brittle in the sense that it contains limited facilities for self-repair. Robustness is typically achieved through use of redundant systems rather than through self-adaptation or repair. Biological systems on the other hand exhibit a remarkable degree of self-adaptation, self-repair and self-reproduction. Unravelling these mechanisms might enable new approaches to engineering of artifacts that display similar properties. This would for example introduce a leap in performance for dependable computing in mission critical applications.