Isolation by distance, web service
© Jensen et al; licensee BioMed Central Ltd. 2005
Received: 31 October 2004
Accepted: 11 March 2005
Published: 11 March 2005
The population genetic pattern known as "isolation by distance" results from spatially limited gene flow and is a commonly observed phenomenon in natural populations. However, few software programs exist for estimating the degree of isolation by distance among populations, and they tend not to be user-friendly.
We have created Isolation by Distance Web Service (IBDWS) a user-friendly web interface for determining patterns of isolation by distance. Using this site, population geneticists can perform a variety of powerful statistical tests including Mantel tests, Reduced Major Axis (RMA) regression analysis, as well as calculate F ST between all pairs of populations and perform basic summary statistics (e.g., heterozygosity). All statistical results, including publication-quality scatter plots in Postscript format, are returned rapidly to the user and can be easily downloaded.
The term "isolation by distance" (IBD) was first used by Sewall Wright [1, 2] to describe patterns of population genetic variation that derive from spatially limited gene flow. IBD is defined as a decrease in the genetic similarity among populations as the geographic distance between them increases. Statistical tests for IBD can be conducted using populations or individuals as the units of replication, although analyses at the individual level typically utilize spatial autocorrelation statistics . When individuals can be grouped into populations, it is possible to calculate 1) a matrix of genetic similarity or distance between all population pairs for comparison with 2) a matrix of pairwise geographic distances. A nonparametric Mantel test is typically used to test for non-random associations between these two matrices. Bohonak  has also suggested performing statistical tests of the IBD slope and intercept using bootstrapped pseudoreplicates, and recently, Yang  has developed a likelihood-based approach for IBD analysis. The ever-growing volume of population genetic data and a continued desire to interpret spatial processes make computational tools for rapidly identifying patterns of isolation by distance highly desirable.
In this paper we describe a sophisticated, yet user-friendly, Common Gateway Interface (CGI) implementation of statistical software for detecting IBD patterns. Isolation by Distance Web Service (IBDWS) is a major expansion and upgrade of Isolation By Distance (IBD), a standalone computer program written in C and originally compiled for Windows and Macintosh . IBDWS includes the following highly useful features for analysis of genetic structure in natural populations:
(1) Analysis of raw data: For a raw data set with codominant markers (e.g., microsatellites), IBDWS calculates basic summary statistics including heterozygosity, F ST between all pairs of populations, and standard transformations of F ST .
(2) Statistical tests: IBDWS performs Mantel tests, partial Mantel tests, reduced major axis (RMA) regression of IBD slope and intercept, and error estimation for slope and intercept using multiple methods.
(3) Simple form interface: Various parameters (e.g., number of randomizations) are easily configured on the IBDWS front page and are checked for validity.
(4) Analysis selection: Multiple analyses may be performed via checkbox selections.
(5) Publication quality graphics: IBD scatter plots are generated in Postscript format for download and further modification.
(6) Rapid analysis: IBDWS is hosted on a fast server with minimal wait times (see Results and discussion).
The majority of these features were not included in IBD. Additionally, IBDWS includes thirty-one updates and bug fixes to the original IBD statistical package. Altogether, these changes make IBDWS a major advancement on the original release in terms of both performance and utility.
IBDWS was written in C++, compiled with the Unix g++ compiler, and runs on a server using Apache 2.0.50 and Fedora Linux. Postscript file plotting capability (see below) was implemented using CGraph, a C++ plotting library. The UNIX command "convert" is called by IBDWS to create a JPEG for web-based viewing from the postscript file. Both files are held on the server for 24 hours until routine deletion.
Different researchers have recommended logarithmic (base 10) transformation of genetic distance, geographic distance or both, prior to IBD analysis [6, 9]. IBDWS automatically performs each of these transformations and provides an interface that switches among the four possible scatter plots.
Results and discussion
Checkboxes and radio buttons allow the user to configure which analyses will be performed (Figure 3). These include the calculation of Rousset's genetic distance  from raw data, options to log-transform genetic and/or geographic distances, replacement values for undefined transformations, and the number of bootstrap randomizations per analysis (up to 30,000).
A full analysis of raw data for six loci, two to three alleles per locus, and 326 individuals required an average wait time of 10 seconds. When an input without any geographic distances is used, wait times are negligible. If raw genotypic data are supplied, IBDWS returns an HTML table with basic summary statistics, the genetic distance F ST between all population pairs, the genetic similarity statistic M  between all population pairs, and Rousset's  distance F ST / (1 - F ST ) if desired. If geographic distances are included, IBDWS will display a scatter plot of genetic distance vs. geographic distance for all pairs of populations, and overlay the RMA regression line (Figure 1). If log-transformation of the data is selected, a pull-down menu below the plot allows the user to choose data sets with one or both axes log-transformed. Clicking on any plot downloads a Postscript version to the user's computer. Statistical analyses of the IBD relationship, including Mantel tests, partial Mantel tests (if appropriate) and the RMA regression analyses described above, are summarized in HTML tables as well. A summary of all analyses can downloaded as a static HTML file, or a tab-delimited text file suitable for importation into a spreadsheet application.
At present, IBDWS requires the user to calculate all pairwise genetic distances between populations for DNA data sets. In future versions, we plan to include algorithms for calculating genetic distances using both sequence and allelic data, including genetic distances based on phylogenetic trees under different models of evolution. When implemented, the user will simply have to supply the raw data and geographic distances before calculations are performed. IBDWS will also be updated with new population genetics analyses as they become available.
In summary, IBDWS is a substantial expansion of the previous PC-based versions of IBD, providing geneticists with a simple interface, multiple statistical calculations, rapid analysis, and high-quality graphics. It will be useful to a significant number of molecular ecology, conservation genetics, population genetics and genomics researchers, and will be updated with newer analyses in the future.
Availability and requirements
IBDWS is currently hosted at http://phage.sdsu.edu/~jensen/. Documentation of the statistical calculations and biological context are available on IBD's original website at http://www.bio.sdsu.edu/pub/andy/IBD.html. The programming source files have been made available at http://sourceforge.net/projects/ibdws/ under the GPL (GNU Public License).
List of abbreviations used
Isolation by distance.
Isolation by distance web service.
Reduce major axis (regression).
The authors are grateful to Forest Rohwer for hosting IDBWS on his web server. We also thank Pat McNairnie for helping us with web configuration, account management and troubleshooting tips. We thank Varykina Thackray for editorial suggestions on the manuscript. Finally, we thank Jay Vavra, Stacey Praizner, and members of the High Tech High School Biotechnology program for arranging the internship that lead to this research project.
- Wright S: Size of population and breeding structure in relation to evolution. Science. 1938, 87: 430-431.Google Scholar
- Wright S: Breeding structure of populations in relation to speciation. American Naturalist. 1940, 74: 232-248. 10.1086/280891.View ArticleGoogle Scholar
- Epperson BK, Li T: Measurement of genetic structure within populations using Moran's spatial autocorrelation statistics. Proc Natl Acad Sci U S A. 1996, 93: 10528-10532. 10.1073/pnas.93.19.10528.PubMed CentralView ArticlePubMedGoogle Scholar
- Bohonak AJ: IBD (Isolation By Distance): a program for analyses of isolation by distance. J Hered. 2002, 93: 153-154. 10.1093/jhered/93.2.153.View ArticlePubMedGoogle Scholar
- Yang RC: A likelihood-based approach to estimating and testing for isolation by distance. Evolution. 2004, 58: 1839-1845.View ArticlePubMedGoogle Scholar
- Slatkin M: Isolation by distance in equilibrium and non-equilibrium populations. Evolution. 1993, 47: 264-279.View ArticleGoogle Scholar
- Rousset F: Genetic differentiation and estimation of gene flow from F-statistics under isolation by distance. Genetics. 1997, 145: 1219-1228.PubMed CentralPubMedGoogle Scholar
- Manly BFJ: Multivariate statistical methods: a primer. 1994, New York: Chapman and Hall, 2Google Scholar
- Hutchinson DW, Templeton AR: Correlation of pairwise genetic and geographic distance measures: inferring the relative influences of gene flow and drift on the distribution of genetic variability. Evolution. 1999, 53: 1898-1914.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.