doi:10.1038/nindia.2012.39 Published online 19 March 2012

Virtual Cell: computing human skin

A computational model of human skin has enormous applications in the cosmetic industry. In future, it may be possible to use virtual skin model to diagnose hair growth, colour and ageing conditions or devise personalized treatment protocols, says Pawan Dhar.

Skin is the only wearable organ of human body. It also happens to be the largest — weighing about 3.6 grams and measuring about 22 square feet in surface area. For an unaided eye, the skin may look like a sheet of cells biochemically glued to the underlying soft tissue. However, detailed study reveals an enormous and complex sensory ecosystem of cellular and molecular interactions that interface muscles, nerves and blood vessels with the external environment.


Human skin is a tightly bound sheet of cells that varies in thickness from 0.5 mm to 5 mm. It houses oil glands, sweat glands, tiny blood vessels and nerve fibres.

Human skin is composed of several layers — the outermost (visible to the unaided eye) is epidermis. Epidermis contains skin pigments. This layer is made of three different sections. The first one is the keratin embedded layer that we shed every day — the stratum corneum. The second section of the epidermis is the keratinocytes layer and the third is the basal layer. The basal layer is where keratin is synthesized.

Just below the epidermis is the dermis that houses blood vessels, lymph vessels, hair follicles and sweat glands held into place by collagen. The subcutis is the lowermost layer of the skin made of fat cells and collagen and thus acts as shock absorber against any injury.

Challenges in modeling human skin

A fully functional computational model would be expected to fulfil at least three things: (1) replicate the overall structure of the human skin in terms of the enormous diversity of its cellular composition (2) virtually model each skin cell type i.e., virtual melanocyte, virtual adipocyte and so on (3) replicate some or all of its basic functions such as sensation, homeostatic regulation to maintain normal temperature, absorption of chemicals, excretion of waste products and oily secretion of the sebaceous glands.

In comparison to the virtual microbe, modelling skin is much more complicated. In contrast to virtual microbe or virtual hepatic cell, in the virtual skin project, one would need to construct several 'virtual cell types' with a hope to make a reasonable integrated model of skin.

For example, one would need to construct virtual melanocyte, virtual keratinocyte and virtual fibroblast to begin with. Second, one would need to connect them by modelling the inter-cellular traffic of molecules. Third, one would need to model processes that connect skin with nervous, muscular, circulatory, lymphatic and bony system.

Probably the first version of computational skin would be a coarse grain model of permeability. This would help cosmetic companies to test their products in-silico purely using permeability criterion. In fact, a computational model based on a set of 61 compounds for Franz cell skin permeation has already been developed1.

The second version of computational skin would be an inter-cellular model that incorporates all the molecular transactions among skin cell types, skin-cell and non-skin cell types. The third version would probably be a comprehensive and conditional molecular connectivity model of individual cell type.

The fourth version would be a semi-quantitative model of human skin incorporating genomics, metabolomics, proteomics, transcriptomics data in each cell type. It is difficult to visualize a completely quantitative human skin model in future, due to data incompleteness and data inaccuracy.

Some key issues that could be resolved by virtual human skin model are: (a) the nature of cell-cell interactions, regulatory, metabolic and cellular components of the whole human skin, (b) the influence of micro and macro environments on the regulatory and signaling processes of individual cells, (c) the emergence of wrinkles through progressive environment-genetic interactions, (d) the impact of intracellular processes e.g., generation of free radicals, on the ageing of skin.

One of the biggest challenges in modeling skin is the enormity of data at the single cell and multi-cellular levels. For example, as of March 2012 there are 423,850 publications listed in Pubmed on human skin, 20,127 papers on human skin fibroblasts, 11,549 papers on human melanocyte and 24,533 papers on human keratinocyte. Considering an overall 50% redundancy of data, we are still left with a morass of papers that needs to be mined for data!

Thus, good literature mining tools are enormously important in capturing patterns from thousands of papers and build a good conceptual template of human skin, to begin with. To collect and integrate enormity of data at several levels, using good literature mining tools combined with manual annotation will be the way to go.

General steps to be followed for constructing virtual cells are: building gene, RNA, protein, interaction inventories, building regulatory (protein and RNA regulation), metabolic and signaling pathways, integrating databases, data mining tools, data prediction, data quality checks, visualization and analysis of pathways, analyzing perturbation and hypothesis generation, developing custom software to take care of requirements from individual cells to virtual tissue.

Progress and future

Last year, a mechanical model of human skin elasticitiy and growth using non-linear mathematical equations was proposed2.The model predicts possible behaviour of human skin beyond its physiological limits. One of the applications of this model is in the field of cosmetic surgery for burn injuries where skin coverage and growth requirements are immense. The model also has potential applications in treatment of reconstructive surgery of congenital conditions.

Another computational model was developed to understand TGF-beta1 signaling process3.It is a three dimensional multiscale model of the human epidermis with potential applications in epidermal wound healing.

Recently, a computational model for the production of (pre-)vitamin D within the skin has been proposed4.This model captures several key physical properties of the human skin.

In future, with more data coming in, a comprehensive virtual human skin is likely to emerge that would be used by clinicians and cosmetic companies. That would truly revolutionize the understanding and applications of human skin.

This article is the sixth in a series entitled 'Virtual Cell'.


  1. Lee, P. H. et al. Development of an in silico model for human skin permeation based on a Franz cell skin permeability assay. Bioorg. Med. Chem. Lett. 20, 69-73 (2010) | Article | PubMed |
  2. Tepole, A. B. et al. Stretching skin:The physiological limit and beyond. Int. J. Nonlin. Mech. doi: 10.1016/j.ijnonlinmec.2011.07.006 (2011)
  3. Adra, S. et al. Development of a Three Dimensional Multiscale Computational Model of the Human Epidermis. PLoS ONE doi: 10.1371/journal.pone.0008511 (2010)
  4. Meinhardt-Wollweber, M. et al. A computational model for previtamin D(3) production in skin. Photochem. Photobiol. Sci. doi: 10.1039/C2PP05295D (2012)