doi:10.1038/nindia.2012.24 Published online 17 February 2012

Virtual cell: Making a virtual human liver

By 2025, a virtual human liver may be used to test new drugs, personalise treatments and significantly reduce animal use in research, Pawan Dhar says.

A virtual liver may be useful in drug testing.
© Pawan Dhar

In the not so distant future, physicians will use genome sequence data and virtual model of cells and organs to genetically profile a drug before recommending it to patients. Indications are that scientists will use virtual liver models to find novel drug targets and pharma companies will use them for screening and development of new drugs.

Why virtual liver?

Liver is a vital human organ that performs a major metabolic role including digestion, storage of glycogen and detoxification. Liver assists in the digestion process by producing almost a litre of bile every day and metabolizing more than 10,000 substances during the same time. These substances consist of food, drugs, alcohol and so on. The body excretes all foreign molecules.

Also, the liver has an amazing capability to regenerate itself almost completely following its damage. Thus, capturing enormous diversity of molecular and cellular connectivities and functions in a factory like setting, are important from the perspective of discovery science and clinical application.

To capture the enormous diversity of molecular processes in a massively parallel liver factory, it is necessary to build a model that provides a virtual ecosystem to understand molecular interactions to network dynamics and whole organ transactions.

The virtual liver is hoped to replicate human liver physiology, morphology and function from molecular layer upwards. Given that liver cells express more genes than other human cell types, one expects to come across a higher deluge in liver system than other cells and organs.

Why systems biology?

Systems biology is an approach to connect a group of interactions with the higher order behavior. The link is made by tying molecular interactions through mathematical equations, simulating the model in computer, going back and forth between model and experiments leading to enhanced understanding of how the system operates at several temporal and spatial scales.

The HepatoSysand the v-Liver programmes aim to integrate cellular level, cell-cell communication, the liver lobule level and the whole organ level data into one computer model.

In 2004, HepatoSys — the largest systems biology consortium making virtual liver cell — was launched in Europe with a 36 million euro grant by the German federal research ministry. By 2010, the consortium had 250 German scientists across 69 research groups in Germany aiming to build a multi-scale model of liver.

The liver modelling group comprises people from biology, chemistry, pharmacology, medicine, theoretical physics, mathematics, computer scientists and engineering backgrounds.

The activities of the HepatoSys network are broadly distributed into three groups — the 'liver cell group' building a mathematical molecular network model of liver cell, the 'beyond the cell group' that focuses on tissue-based model and the 'multi-scale modelling group' that will integrate all cell-to-tissue models and translate it towards clinical applications.

HepatoSys comprises several interconnected projects. One studies the formation and transport of vesicles within the cell and their impact on the signaling processes while another looks at iron metabolism and modeling of complex regulatory systems that controls absorption, distribution and excretion of the trace element. One project probes how the liver regenerates so efficiently and another studies reconstruction and analysis of liver cell metabolism under different conditions.

The virtual liver project seems to have gained momentum in the US also. Last year, the Environmental Protection Agency (EPA) in United States started a virtual liver project (v-Liver) with a funding of US$ 3 million. The project aims at predicting long term exposure to small quantities of chemicals in food and water, and estimate the potential of chemicals to cause chronic diseases such as cancer using computer models and simulation.


For experimental research, the HepatoSys group uses liver extracts from freshly dissected animals rather than liver cell lines since the cell lines are known to lose their original characteristics over time. It took several years for the consortium to build standard operating procedures to ensure that everyone followed the same protocol and the results were comparable across the labs.

Model building is a slow process of carefully hand-weaving molecular reactions into a network. It took five years to develop a model to simulate the flow of nutrients and signalling molecules into the cell. Groups working to understand metabolic network of human hepatocytes have succeeded in integrating a few hundred enzyme-catalysed reactions and more than 400 metabolites into the model system.

In the first phase, the focus was to understand stationary metabolic fluxes in the central metabolism of hepatocytes. In the second phase, metabolic dynamics of the hepatocytes will be studied in response to cholesterol-lowering drugs.


Even if accurately built over time, the animal cell liver model isn't close to human liver metabolism, gene regulatory and signalling system. Due to this reason, findings from rat liver cell have to be extrapolated to human system, in particular the toxicity data that cannot have huge error margins.

Another constraint is that closely bred lab animals are genetically identical in contrast to humans, who are genetically quite diverse. Here again, one will have to do quite a bit of extrapolation.

Liver cell models will continue to be built from cell to tissue, using various logical and mathematical approaches. Once built, the speed of the computer will be the rate limiting step in testing a potentially harmful chemical on a computer model of a human liver. Such a paradigm shift would be transformational, both for academia and for industry.