InvFEST, a database integrating information of polymorphic inversions in the human genome

Abstract

The newest genomic advances have uncovered an unprecedented degree of structural variation throughout genomes, with great amounts of data accumulating rapidly. Here we introduce InvFEST (http://invfestdb.uab.cat), a database combining multiple sources of information to generate a complete catalogue of non-redundant human polymorphic inversions. Due to the complexity of this type of changes and the underlying high false-positive discovery rate, it is necessary to integrate all the available data to get a reliable estimate of the real number of inversions. InvFEST automatically merges predictions into different inversions, refines the breakpoint locations, and finds associations with genes and segmental duplications. In addition, it includes data on experimental validation, population frequency, functional effects and evolutionary history. All this information is readily accessible through a complete and user-friendly web report for each inversion. In its current version, InvFEST combines information from 34 different studies and contains 1092 candidate inversions, which are categorized based on internal scores and manual curation. Therefore, InvFEST aims to represent the most reliable set of human inversions and become a central repository to share information, guide future studies and contribute to the analysis of the functional and evolutionary impact of inversions on the human genome.

Figure 1.

(A) Diagram of the InvFEST data model and processing. The dotted box shows a simplified star-like schema of the InvFEST database. The information processed by the automatic InvFEST merging engine is shown in blue and connected by arrows, while the process of manual addition of validations and other data is shown in green and connected by dashed lines. (B) Automatic definition of inversion breakpoints through the InvFEST merging engine. Assigned breakpoints correspond to the overlap between the breakpoints of all individual predictions, always taking into account the resolution of each study methodology (shown in grey).

Figure 2.

Summary of the InvFEST database content. (A) Status of the 1092 InvFEST candidate inversions. Numbers in parentheses indicate number of inversions for each status category. (B) Overlap among the predictions coming from different studies (with reference indicated in parentheses). Numbers of inversions predicted by one single study are shown in red, while black numbers indicate number of inversions supported by two or more studies. Small Venn diagram shows the overlap between the 22 inversions identified by particular studies and 1089 genome-wide predictions. See an interactive version of this figure in the Supplementary Data. (C) Number of inversions supported by 1, 2, 3, 4, 5 or 6 different studies. Different status categories are shown in colors and its percentage is represented in the table.