InvFEST: Human Polymorphic Inversion DataBase

InvFEST User's Guide

Table of Contents:

Where do InvFEST inversions and related information come from?
How are inversions defined in InvFEST?
What filters have been applied to the inversions catalogued in InvFEST?
What do different inversion status mean?
What information is included in an inversion report?
- General information
- Region map
- Predictions
- Validation and genotyping
- Frequency
- Breakpoints
- Evolutionary history
- Functional effects
- Report history
What do different scores by Martinez-Fundichely et al. (2013, in preparation) mean?
Why do we not translate inversion coordinates into hg19?
What is the data model of InvFEST?

+ Where do InvFEST inversions and related information come from?

+ How are inversions defined in InvFEST?

InvFEST inversions are a non-redundant dataset of inversions generated by overlapping the position of the breakpoints of each individual prediction as reported in the original publication and taking into account the resolution of the study methodology.

The breakpoints reported in InvFEST for each inversion may have been automatically or manually defined. Then, the parameter “Definition method” may be one of the following:

Default informatic definition . The breakpoints reported are automatically computed by the InvFEST engine (OLAP), either when the inversion is initially generated or because the inversion is automatically modified by adding a new inversion prediction that merges into it. This informatic definition defines each breakpoint location as the maximum overlap between the breakpoints of all individual predictions, always taking into account the resolution of each study methodology.
Manual curation . Breakpoints location for inversions might be manually curated after their inicial default informatic definition. These manually curated breakpoints are the most accurate breakpoint locations for an inversion and are defined through a validation assay.

It is important to keep in mind that inversions in the InvFEST database are defined relative to the NCBI Build 36.1 (hg18) human genome reference assembly (produced by the International Human Genome Sequencing Consortium on March 2006), and for simplicity the standard orientation is always the same as the reference and the inverted is the opposite one, independently of which one might be ancestral. Please read more information about reference assemblies and coordinate conversion in the section Why do we not translate inversion coordinates into hg19?.

+ What filters have been applied to the inversions catalogued in InvFEST?

+ What do different inversion status mean?

The status of InvFEST inversions indicate their fiability, according to different bioinformatic and/or experimental results:

Validated : At least one breakpoint of the inversion has been validated experimentally according to the published information.
Predicted : These inversions have not been experimentally checked and have simply been predicted by one or more high-throughput methods.
Unreliable prediction : The inversions have not been experimentally checked, and all their predictions either do not pass the internal bioinformatic quality criteria of their own study [such as the set of scores in the GRIAL algorithm (Martínez-Fundichely et al., in preparation)], or have their breakpoints overlapping more than 90% of their length with simple repeats, low complexity repeats, or satellite repeats identified by RepeatMasker (since the presence of this kind of repeats tends to generate unreliable PEM predictions from short reads generated by next generation sequencing).
Ambiguous : This is a temporary status automatically assigned by InvFEST if the results of two or more validation assays are contradictory.
False : The inversion has been invalidated experimentally or the predictions that supported the inversions are incorrect.
Obsolete : Former versions of inversions that have been manually joined or split into new inversions, and that do not appear in the inversion list anymore (although they can be searched specifically by InvFEST identifier).

+ What information is included in an inversion report?

Each inversion report contains the following sections:

General information: this first section is a summary of the whole report, including e.g. the accession code, the region of the inversion, the estimated inversion size (i.e. length of the inverted segment from the middle position of the two breakpoint intervals), the global inverted allele frequency, or the mechanism of origin.
Region map: overview of the inversion genome region, including genes, segmental duplications, InvFEST inversion, and corresponding predictions. The image is a link to the same region at the UCSC Genome Browser.
Predictions: all the individual predictions are reported in this section, including e.g. a brief description of the study, the breakpoints location, or the individuals of the study on which the inversion was predicted. The title of each subsection is a link to Pubmed for published studies.
Validation and genotyping: results of experimental validations, including e.g. the validation method, the genotyping results, and the resulting status. The title of each subsection is a link to Pubmed for published studies.
Frequency: population data for each continent and population analyzed, including e.g. inversion frequency and fit to the Hardy-Weinberg equilibrium. There is also an utility to generate frequency graphs for continents or specific populations.
Breakpoints: information regarding the breakpoints, including e.g. the genome position, definition method (automatic or manually curated), mechanism of origin, and sequence features (e.g. segmental duplications).
Evolutionary history: information regarding the orientation of the inversion in other species and estimated ancestral orientation, and the estimated age and origin of the inversion.
Functional effects: genes within or close to the inversion that might be affected. Reported information includes e.g. the mechanism for which the inversion disrupts the gene and the functional consequences. The title of each subsection is a link to NCBI Genes.
Report history: this section reports any manual curation performed on the inversion breakpoints since its initial automatic definition. When an inversion replaces former obsolete inversions, this is also reported here.

+ What do different scores by Martinez-Fundichely et al. (2013, in preparation) mean?

The definition of inversions using paired-end mapping (PEM) based on a pre-decided support (minimum number of paired-ends (PEs) supporting the inversion) introduces many false positives to the final set of predicted inversions mainly due to mapping errors, errors of the PEM technique, or other biological phenomena that may result in inversion-like PEM patterns. Martinez-Fundichely et al. (2013, in preparation) tries to overcome this problem by applying different scores to the set of predicted inversions. We include the results of two of these in InvFEST:

D/C Test : statistical test applied to the ratio of the numbers of Discordant (D) to Concordant (C) PEs. Given that the number of concordant PEs denying the inversion is not expected to exceed the number of discordant PEs supporting it (even if the inversion is homozygote or heterozygote), a score is applied trying to quantify this. When this score is equal or lower than -1.96, the inversion does not pass the test and it is categorized with the status Unreliable prediction. Scores not exceeding the minimum threshold are highlighted in red.
Disc. Mapping Prob. : this test tries to quantify what is the probability that an inversion of a given size is supported by less or equal the observed number of discordant PEs. The test corrects the results based on the number, distribution, and percentage of identity of repeated sequences in the breakpoints (a kind of mappability correction). If the probability is lower than a threshold (<0.001), the inversion is categorized with the status Unreliable prediction. Scores not exceeding the minimum threshold are highlighted in red.

Inversions in InvFEST that have been detected by GRIAL (Martinez-Fundichely et al. 2013, in preparation) show the results of these two scores in the 'Predictions' section.

+ Why do we not translate inversion coordinates into hg19?

+ What is the data model of InvFEST?

InvFEST is a data-warehouse implementation, commonly used to start schema data models. The center of the star is the “inversion” table, which is de-normalized for speed purposes. The points of the star are the dimension tables that contain related information (e.g. predictions, validations, evolutionary information, etc.). Below follows a detailed Entity-Relationship (ER) schema of InvFEST and a list enumerating all the tables of the database with a description and all relevant areas of the InvFEST processing engine they are involved in.

Table Name	Description	Relevant Areas of the InvFEST Processing Engine
*breakpoints*	Second central table of the database. Contains a summary of all basic information about all inversion breakpoints in the database (chromosome location, definition date, source, description, association with segmental duplications, association with genes, general comments, validation ID). Includes all defined breakpoints for all inversions during all the history of the database; those displayed as final breakpoints of the inversion in the inversion report are the latest manually defined breakpoints in the table if available, or the latest defined breakpoints in the table otherwise	Predictions , Validations , Merging Engine
*fosmids*	Stores codes and accession numbers of fosmids used by predictions and/or validations	Predictions , Validations
*fosmids_predictions*	Linking table between tables predictions and fosmids (which fosmids are supporting each prediction, if applicable)	Predictions
*fosmids_validation*	Linking table between tables validation and fosmids (which fosmids are supporting each validation, if applicable)	Validations
*genomic_effect*	Association with genes ( RefSeq genes annotated on hg18) for each defined breakpoints of the database ( bioinformatic prediction of how the breakpoints affect the gene, observed functional effects on the gene, source). Automatic associations included in the table are conservative: only shown if the gene sequence is disrupted by the inversion (not if the complete gene is inverted or is close to the inversion), and only shown if no transcripts of a gene are left unaffected; when affected, if different transcripts are affected differently, only less drastic effect is described	Predictions , Validations , Associated information , Merging Engine
*HsRefSeqGenes*	RefSeq genes on hg18 as downloaded from the UCSC	Predictions , Merging Engine
*HsRefSeqGenes_exons*	Exons extracted from the HsRefSeqGenes table	Predictions , Merging Engine
*HsRefSeqGenes_introns*	Introns extracted from the HsRefSeqGenes table	Predictions , Merging Engine
*individual_research*	Individual codes analyzed in specific studies	Predictions , Validations
*individuals*	Individual codes with associated information (nickname, gender, population, family, relationship in the family, panel)	Associated information
*individuals_detection*	Links IDs of individuals to specific predictions and/or validations	Predictions , Validations
*inv_age*	Estimated age for inversions (source, method, estimated age)	Validations , Associated information
*inv_origin*	Evolutionary origin of inversions (source, method, unique/recurrent)	Validations , Associated information
*inversion_history*	History report of inversions manually merged or split in the database (previous inversion ID, new inversion ID, cause)	Validations
*inversions*	Central table of the database. Contains a summary of all basic information about all inversions in the database (name, chromosome location, size, frequency, number of predictions, number of validations, origin, status)	Predictions , Validations , Associated information , Merging Engine
*inversions_in_species*	Orientation of the inversion in species other than humans (species ID, source, method, orientation: standard=equal to hg18 reference assembly / inverted=opposite to hg18 reference assembly)	Validations , Associated information
*phenotipic_effect*	Observed phenotypic effects of inversions in the database (source, mechanism, observed effects)	Validations , Associated information
*population*	Population name and region name	Associated information
*population_distribution*	Stores population distribution information (number of analyzed individuals, number of inverted alleles, population frequency), for specific populations obtained by specific validations	Validations , Associated information
*predictions*	Stores all predictions from individual studies (including coordinates, status, bioinformatic scores), and links to tables inversions (there is redundancy in the predictions table because the relationship is m:n ) and researchs	Predictions , Merging Engine
*researchs*	Information about research articles from which InvFESTdb information is extracted (name, description, citation, number of individuals analyzed, prediction method, prediction error, validation method)	Predictions , Validations , Associated information
*SD_in_BP*	Association with segmental duplications (as downloaded from the UCSC) for each defined breakpoints of the database. Also describes if the same segmental duplication affects one or both breakpoints, and in the latter case, if the segmental duplication is in direct or indirect orientation in one breakpoint versus the other	Predictions , Merging Engine
*seg_dups*	Segmental duplications onhg18 as downloaded from the UCSC	Predictions , Merging Engine
*species*	List of species other than humans (ID, species common name)	Associated information
*validation*	Stores all validations from individual studies (including method, experimental conditions, status), and links to table researchs	Validations