Notes on the Animal QTL Database Development

Initial proposals, plans, and design scratches on developing a (pig) QTL database was made within the NAGRP Pig Genome Coordination Program in the fall of 1998. During the early days of the initial developments, much of the informal discussions about visions, designs and implementations were among scientists from Roslin Institute (UK) and within Iowa State University. Part of the missions of my 2-week working visit to the Roslin in the summer of 1999 was on this. Those who were involved in the discussions include Drs. Alan Archibald, Jian Hu, Andy Law, Lizhen Wang, Jack Dekkers, et al. Discussions with John Bastiaansen on trait data organizations was also useful while I served as an consultant to PIC in late 2001. The database layout, design, initial data curation and preliminary exploration on the approaches implementing user portals were actually tried out in 2002/2003 when I returned to Iowa State University where I stayed as an Associate Scientist until now.

Working with NCBI team was a boost to the QTLdb development in terms of implementation of a data structure using their mouse QTL data template. Under an agreement between the NAGRP Bioinformatics Coordinator and the NCBI, a copy of the database is implemented at the NCBI. Thanks for the works of Drs. Donna Maglott, Svetlana Dracheva, and Wonhee Jang, a preliminary release with partial data was made on NCBI portals in June 2004 (Release 1). Subsequently the NAGRP made a second release (Release 2) with many customized user interface tools in December, 2004 as the data collection and new tools development came to a completion. As initially planned, the pig QTL data is synchronized between the NCBI and NAGRP databases. During the re-development of Locus Link into its successor Gene DB, the QTL data synchronization was at a pause during 2006-2009 and was resumed in 2009, with continued efforts in making the data stream smooth.

The pig QTL data visualization tools were developed in 2004, which resulted in our first paper on PigQTLdb published by Mammalian Genome in 2005 (Volume 16(10):792-800). subsequently, the QTLdb was expanded for new functionalities and utilities. The first of which was the addition of the set of curator/editor tools. The second of which was the expansion of the QTLdb to include cattle/chicken QTL data in 2005/2006. The third of which was additions of structural alignment of the QTL maps with RH maps, human maps, FPC maps etc., with includsion of SNPs, Microarray elements, and new microsatellites for alignments in 2006/7 (see FAQ #11 for details). As a result, two more papers were published ( Nucleic Acids Research, 2007, 35 [Database issue]: D604-D609; Mammalian Genome, 18: 1-4, 2007 on a new database name, Animal QTLdb).

During the course of the project development, many attempts were made to utilize existing tools. Initially, we contacted a number of research groups and attempted several existing tools, which include: Roslin Institute for Anubis¹ tool, Cold Spring Harbor Lab for cMAP, University of Tennessee Bioinformatics group for their Mouse/Human Bone Density QTL Database², Sweden RatMap³ Group for its QTL tools, Texas A&M University for a Bovine QTL Viewer, among others. For various reasons none of these tools was worked out with the available facilities, against our list of goals in terms of both wanted functions and utilities. As the last resort we explored a Perl/GD and MySQL approach to program out the needed tools for integrating, displaying and analyzing the stored QTL data. All current peripheral tools and utilities as they stand now are built on this basis.

Releases of the QTLdb with new data and functions were made at lease once per year in the earlier years (2004-2008), and was made 3 times a year (since 2009) as we took more aggressive approaches by implementing multiple channels for new data curation, and dedicated curator for routine data curation activities. New additions of functions and tools became as part of the routine as well (See Animal QTLdb Releases).

The database and its peripheral tools at the NAGRP implementation was designed mainly for users to compare, confirm and locate the most plausible locations on a chromosomes for QTL. This is aimed at providing a tool for structural genomic information mining for genes responsible for quantitative traits important to animal production; The NCBI implementation has all marker information matched to marker records in NCBI's UniSTS database. This allows automatic matching of markers to public sequence data by e-PCR. Data on the NCBI and the NAGRP Animal Genome servers are cross-referenced to each other. This function allows all of the unique information on each site to appear to be integrated on the same database server via worldwide web.

A noticible extended feature developed into the QTLdb is that diverse types of structural genome features, such as microsatellites, SNPs, microarray elements, tiled FPC BACs, can be aligned with QTL map locations to aid the data mining. Further more, different map types, such as RH maps, genome sequence maps, human maps, can also be align where available data is available to establish links. This functionality is further enhanced with the use of GBrowse tool for alignment of essentially any structural genomic features, such as transcripts, mRNA and gene structures mapped to the genome.

As part of our efforts to benefit the research community, we offered the QTLdb software tools to Jill Maddox of The University of Melbourne to set up a Sheep QTLdb in Australia in March, 2007. The Sheep QTL data were curated, and the QTLdb maintain, by Jill Maddox until 2010 when the site was migrated back to Iowa State University in October, 2010.

In order to cope with increasing demands on daily database queries, a new development was to establish mirror sites to distribute the server and network loads. The first Animal QTLdb mirror site was set up in the summer of 2013 in Wuhan, China with a courtesy by Dr. Shu-hong Zhao's group at Huazhong Agriculture University, Hubei, China. Network traffic from 18 Asia countries are currently redirected to this site: http://cn.animalgenome.org/QTLdb (The mirror site ceased operation after 2014)

Further developments of the database in the 2010s were geared towards extentions of QTLdb under a federated database model. A good example is the successful integrated development of the Animal Trait Genetic Correlation Database (CorrDB) by sharing common resources and tools in the backend, such as trait management with integrated trait ontology developments (see Vertebrate Trait (VT) Ontology, Livestock Product Trait (LPT) Ontology, Clinical Trait Ontology, and Livestock Breed Ontology). Other featured developments include additions of new species (rainbow trout, catfish, sheep, horse, etc) and new data types (GWAS, CNV, Haplotypes, epistatics, pleitropics, gene networks, polygenic data links, etc), new data curation/edit tools (automations of new data imports and managements from Pubmed, batch curation tools, permanent data locator to facilitate publication reviews, new trait data types with modifiers, etc.), new users tools such as genome alignment tools with GBrowse and JBrowse, gene and trait centric views of data, and establishing external data alliances with Ensembl, UCSC, Reuters Data Citation Index, Monarch Project, etc (previously we already established data flows with NCBI GeneDB). etc.

The success of our efforts developing the QTLdb is marked with a series of publications throughout the course of this work:

In total, our publications on developing the QTLdb have been cited 1,989 times⁵ as of August 27, 2024. In addition, the use of Animal QTLdb has been mentioned 3,570 times as of August 27, 2024 according to Google Scholar.

1. Anubis: A graphic tool made in Java that draws (comparative) linkage maps. It was developed by Chris Mungal at the Roslin Institute and served the Arkdb during 1995-2003. Now it's no longer supported.
2. Mouse/Human Bone Density QTL Database: URL: http://132.192.64.52/qtl/ by Weikuan Gu's group at the University of Tennessee. It is no longer publicly available.
3. RatMap QTL Tools: URL: http://ratmap.gen.gu.se/ by Sweden RatMap Group. No longer publicly available.
4. Sheep QTLdb: The Sheep QTL data was curated and maintained at the The University of Melbourne (Australia) between March 2007 and November 2010 by Jill Madox using the QTLdb software suite. The Sheep QTLdb was migrated from its Australia to its Iowa State University site in October 2010.
5. This number may be inflated because a publication may cite more than one of our papers on the QTLdb.