An article in the open-access journal PLOS Computational Biology discusses how statistical modeling can aid researchers in determining a person’s individual immune system proteins. This will help to assess a patient’s disease fighting ability or the likelihood that the immune system will attack the body’s own tissues.

Although many areas of research and medicine rely critically upon this knowledge, obtaining this information is usually costly and difficult. The study, authored by J. Listgarten and colleagues, shows that information on immune system proteins can be acquired more easily and cost-effectively.

The human immune response is based on specialized immune cells that are sensitized to see small pieces of foreign pathogens, such as viruses. These cells are part of the immune system’s “train-to-kill” mechanism. After the sensitization, these cells are then activated to kill cells that present the same piece of pathogen. The pathogen must join together with the person’s specialized immune proteins, or HLA (human leukocyte antigen), in order for the sensitization and killing to occur.

The human immune response is dependent upon the way in which pathogen peptides (chains of amino acids) interact with these HLA proteins. If we know which HLA proteins that a person has, the information can be used to improve transplant medicine, to find immunogenetic risk factors for disease, and to understand how viruses like HIV change inside the body and continue to escape the immune system.

The model presented by Listgarten and colleagues finds statistical patterns in a large set of previously measured, high-quality HLA data. The research team uses the patterns to improve low-quality HLA data so that it is of higher quality and more informative than the original measured laboratory data. The tool is public and available to the research community, helping analysts improve the quality of their HLA data and improving the study of individual immune systems.

Statistical Resolution of Ambiguous HLA Typing Data
Listgarten J, Brumme Z, Kadie C, Xiaojiang G, Walker B, et al. (February 2008)
PLoS Computational Biology 4(2): e1000016.
doi:10.1371/journal.pcbi.1000016
Click Here to View Journal Website

Written by: Peter M Crosta