The H-Invitational Database (H-InvDB), a comprehensive annotation resource for human genes and transcripts

Abstract

Here we report the new features and improvements in our latest release of the H-Invitational Database (H-InvDB; http://www.h-invitational.jp/), a comprehensive annotation resource for human genes and transcripts. H-InvDB, originally developed as an integrated database of the human transcriptome based on extensive annotation of large sets of full-length cDNA (FLcDNA) clones, now provides annotation for 120 558 human mRNAs extracted from the International Nucleotide Sequence Databases (INSD), in addition to 54 978 human FLcDNAs, in the latest release H-InvDB_4.6. We mapped those human transcripts onto the human genome sequences (NCBI build 36.1) and determined 34 699 human gene clusters, which could define 34 057 (98.1%) protein-coding and 642 (1.9%) non-protein-coding loci; 858 (2.5%) transcribed loci overlapped with predicted pseudogenes. For all these transcripts and genes, we provide comprehensive annotation including gene structures, gene functions, alternative splicing variants, functional non-protein-coding RNAs, functional domains, predicted sub cellular localizations, metabolic pathways, predictions of protein 3D structure, mapping of SNPs and microsatellite repeat motifs, co-localization with orphan diseases, gene expression profiles, orthologous genes, protein–protein interactions (PPI) and annotation for gene families. The current H-InvDB annotation resources consist of two main views: Transcript view and Locus view and eight sub-databases: the DiseaseInfo Viewer, H-ANGEL, the Clustering Viewer, G-integra, the TOPO Viewer, Evola, the PPI view and the Gene family/group.

INTRODUCTION

Human transcripts represent a biologically and functionally rich format for examining the structure of human genes and alternative splicing isoforms. In particular, cloning and sequencing of full-length cDNAs (FLcDNAs) that cover all exons but no introns can facilitate the precise determination of human gene structure (1). Studies on human transcripts have thus been systematically and extensively carried out to draw the outline of the human transcriptome (2–6). The human transcriptome consists of protein-coding mRNAs and non-coding functional RNAs. Analysis of these sequences will provide insights into how genomic information is transformed into higher order biological phenomena. By comparative analysis of the transcriptome with the human genome, we will be able to determine the transcribed regions of the genome and better understand the regulatory machinery of transcription (7, 8). It is therefore of great significance to collect information about human transcripts as well as their annotations. We thus held the first international workshop entitled ‘Human Full-length cDNA Annotation Invitational’ (abbreviated as H-Invitational or H-Inv) in Tokyo, Japan from 25th August to 3rd September 2002, and constructed a novel, integrative database of the human transcriptome, called H-InvDB (9,10). This consists of the annotation of 42 421 human FLcDNAs, collected from six high-throughput producers of human FLcDNAs in the world human gene collections.

To cover the increased number of human FLcDNAs since the initial release of H-InvDB, we held the second international annotation meeting entitled ‘H-Invitational 2 Functional Annotation Jamboree’ (abbreviated as H-Invitational 2 or H-Inv2) in Tokyo, Japan from 15th to 20th November 2003. The second major release of H-InvDB (release 2.0) was based on the annotation carried out at the H-Inv2 annotation jamboree. After H-Inv2, we initiated the Genome Information Integration Project (GIIP) and held the third and fourth annotation meetings in October 2005 and October 2006. The products of those two annotation meetings comprised releases 3.0 and 4.0 of H-InvDB. The increases in the number of entries in H-InvDB are summarized in Table 1.

Table 1.

Statistics of H-InvDB entries

H-InvDB release	Date of release	Number of transcripts (HIT)	Number of gene clusters (HIX)	Number of proteins (HIP)	Human genome	Date of sequence data-fix
1.0	2004/4/20	41 118	21 037	–	NCBI build 34.1	2002/7/15
2.0	2005/8/31	56 419	25 585	–	NCBI build 34.1	2003/9/1
3.0	2006/3/31	167 992	35 005	–	NCBI build 35.1	2005/3/1
4.0	2007/3/30	175 542	34 701	116 228	NCBI build 36.1	2006/6/15
4.6	2007/9/27	175 536	34 699	116 142	NCBI build 36.1	2006/6/15

THE ANNOTATION IN OUR LATEST UPDATE, H-InvDB 2007

In our latest release H-InvDB_4.6, we annotated 120 558 human mRNAs extracted from the International Nucleotide Sequence Databases (INSD) in addition to 54 978 human FLcDNAs that were available on 15th June 2006. We mapped those human transcripts onto the human genome sequences (NCBI build 36.1) and determined 34 699 human gene clusters, which could define 34 057 (98.1%) protein-coding and 643 (1.9%) non-protein-coding loci, while 858 (2.5%) transcribed loci overlapped with predicted pseudogenes. We basically followed the mapping technique we described previously (9,10). We updated annotation for the mitochondrial transcripts since the previous major release, H-InvDB_4.0, which resulted in a slightly decreased number for the transcripts and clusters. Then we assigned a standardized functional annotation to each H-Inv transcript by human curation, based on the results of similarity searches and InterProScan (11). The numbers of manually curated human proteins in each category are summarized in Table 2.

Table 2.

Statistics of manually curated representative H-Inv proteins

Category	Definition	Number of representative HITs	%
I	Identical to knowna human protein (≥98% identity, =100% coverage)	12 404	36.42
II	Similar to knowna protein (≥50% identity, ≥50% coverage)	3165	9.29
III	InterPro domain containing protein	3056	8.97
IV	Conserved hypothetical protein	4210	12.33
V	Hypothetical protein	5124	15.05
VI	Hypothetical short protein (20–79 amino acids)	5250	15.42
VII	Pseudogene candidates	858	2.52
Total		34 057	100

^a‘Known’ proteins are experimentally validated proteins in literatures.

For these transcripts and genes, we provide comprehensive annotation including descriptions of their gene structures, alternative splicing isoforms, functional non-protein-coding RNAs, functional domains of proteins, predicted sub cellular localizations, metabolic pathways, predictions of protein 3D structure, mapping of SNPs and microsatellite repeat motifs, co-localization with orphan diseases, gene-expression profiles, orthologous genes and evolutionary features in model animals, protein–protein interaction (PPI) and annotation for gene families. We have also annotated several new features related to transcript quality.

NEW ANNOTATED FEATURES IN H-InvDB

Classification of ncRNA

We annotated the transcripts that do not have homology to known protein-coding genes or InterPro-domain-containing genes as non-protein-coding transcript candidates. We classified 1216 non-protein-coding transcripts into ‘Identical to known ncRNA’ (124), ‘Similar to known ncRNA’ (74) and ‘Putative ncRNA’ (1018) by homology with known ncRNA databases and discrimination analysis

Sequence quality features: nonsense-mediated decay (NMD), read-through, reverse orientation

A total of 269 transcripts were annotated as candidates of read-through and 2731 as targets of NMD by the extended sequence quality annotation.

Category VII: pseudogene candidates

To annotate transcribed pseudogene candidates, we did the following: First, we filtered out the functional protein-coding genes by only targeting representative category II transcripts and those identified to have frame shifts and/or nonsense mutations; Second, we predicted transcribed pseudogene candidates based on a support vector machine (SVM) method. In the current release, we annotated 1112 transcribed pseudogene candidates (Category VII).

Annotation of gene families/groups

We annotated four selected gene families/groups: T-cell receptor (TCR), Immunoglobulin (Ig), Major Histocompatibility Complex (MHC) or Human Leukocyte Antigen (HLA) and Olfactory receptor (OR) using the original pipeline based on sequence analysis against genome and protein databases complemented by a text-mining approach. In the current release, we identified 15 TCR, 21 Ig, 72 MHC and 122 OR gene clusters.

All the annotation items and features of H-Inv transcript sequences are stored and shown in the main views or sub-databases in H-InvDB.

COMPREHENSIVE ANNOTATION RESOURCES IN H-InvDB

The current H-InvDB annotation resources consist of two main views, Transcript view and Locus view, and eight sub-databases: the DiseaseInfo Viewer, H-ANGEL, the Clustering Viewer, G-integra, the TOPO Viewer, Evola, the PPI view and the Gene family/group view with the appropriate cross-links. An overview of the comprehensive annotation resources of the human gene and transcripts in H-InvDB is shown in Figure 1.

Figure 1.

H-InvDB: overview of the comprehensive annotation resource for the human genes and transcripts. The current H-InvDB annotation resources consist of two main views, Transcript view and Locus view, and eight sub-databases: the DiseaseInfo Viewer, H-ANGEL, the Clustering Viewer, G-integra, the TOPO Viewer, Evola, the PPI view and the Gene family/group view. The Transcript view and the Locus view are the main viewers to display the annotation of each H-Invitational transcript (HIT) and H-Invitational cluster (HIX). The DiseaseInfo Viewer, H-ANGEL, the Clustering Viewer, G-integra, the TOPO Viewer, Evola, the PPI view and the Gene family/group view are sub-databases to provide detailed annotation for each annotation feature. The links to related databases are provided from the appropriate viewers.

Transcript view

The transcript view shows all the annotation of the H-Inv transcript in 12 section tabs: (i) gene structure, (ii) gene function, (iii) gene ontology, (iv) predicted CDS, (v) functional motif, (vi) sub cellular localization, (vii) protein structure information, (viii) gene expression, (ix) disease/pathology, (x) evolutionary information, (xi) polymorphism (SNP, indel and microsatellite) and interspersed repeat information and (xii) transcript and sequence quality information. As seen in the example of a transcript view shown in Figure 1, this view also has links to many external public databases including DDBJ/EMBL/GenBank, RefSeq, UniProtKB, HGNC, InterPro, Ensembl, EntrezGene, PubMed, dbSNP, GO and GTOP and to web sites of the original data producers of the FLcDNA clones and sequences including the Chinese National Human Genome Center (CHGC), German cDNA Consortium (DKFZ/MIPS), Helix Research Institute, Inc. (HRI), the Institute of Medical Science in the University of Tokyo (IMSUT), the Kazusa DNA Research Institute (KDRI), the Mammalian Gene Collection (MGC/NCI) and NEDO. This view was previously known as the cDNA view (mRNA view).

Locus view

The Locus view shows all the annotation of a locus in six section tabs: (i) gene structure and location in the human genome, (ii) gene function, (iii) alternative splicing pattern, (iv) gene expression, (v) disease/pathology and (vi) cluster member information. As seen in the example of a Locus view shown in Figure 1, it shows links to external public databases including DDBJ/EMBL/GenBank, RefSeq, EntrezGene, GeneCards, HGNC and OMIM.

DiseaseInfo Viewer

The DiseaseInfo Viewer is a database of known and orphan genetic diseases and their relation to H-Inv clusters with EntrezGene and OMIM cross-links. The DiseaseInfo Viewer provides two kinds of disease information related to H-Inv clusters: known disease-related genes and co-localized orphan diseases. An orphan disease is defined as a disease mapped on a chromosomal region, but for which the responsible gene has not been identified yet. Co-localization does not necessarily mean a direct relationship between gene and disease; however, genes that are cytogenetically co-localized with a disease could be possible candidate genes for that disease. The co-localized H-Inv clusters are chosen by computing the physical range of each cytogenetic band with a 1 Mbp margin.

Human anatomic gene expression library (H-ANGEL)

H-ANGEL is a database of expression patterns that we constructed to obtain a broad outline of such patterns for human genes (12). We collected gene-expression data in normal and adult human tissues that were generated by three types of methods and in seven different platforms, including: iAFLP, a PCR-based quantitative expression profiling method; DNA arrays (long oligomers, short oligomers and cDNA microarrays); and cDNA sequence tags (SAGE, EST, BodyMap and MPSS). The H-ANGEL database comprises the largest and most comprehensive collection of gene expression patterns so far, which also provides a classification of human genes in terms of their expression.

Clustering Viewer

The Clustering Viewer facilitates the comparisons of different clustering. It allows users to see whether H-Inv transcripts are consistently clustered by different clustering methods. It also displays multiple alignments of transcripts by using CLUSTALW (13). The Clustering Viewer shows all the member transcripts of an H-Inv cluster to which a query sequence belongs.

G-integra

G-integra is an integrated genome browser, in which we can examine the genomic structures of the transcripts. As seen in an example view in Figure 1, the location in the human genome and gene structure of H-Inv transcript (green), and the corresponding RefSeq and Ensembl entries are shown. The structures of the genes and transcripts for 11 non-human species, Pan troglodytes (chimpanzee), Macaca sp. (macaque), Mus musculus (mouse), Rattus norvegicus (rat), Canis familiaris (dog), Bos taurus (cow), Monodelphis domestica (opossum), Gallus gallus (chicken), Danio rerio (zebrafish), Tetraodon nigroviridis (tetraodon) and Takifugu rubripes (fugu) can be optionally displayed for comparison. Other options allow the, the results of gene prediction programs such as GenScan (14), HMMgene (15), FGENESH (16) and JIGSAW (17) to be displayed.

TOPO Viewer

The TOPO Viewer is a tool for viewing subcellular targeting signals predicted by TargetP (18) and the presence of transmembrane helices predicted by SOSUI (19) and TMHMM(20). The probabilities that a protein may be delivered to up to nine distinct sub cellular locations are predicted by WoLF PSORT (21). TargetP predicts whether a protein contains a signal peptide, a mitochondrial targeting signal or any other type of signal. The TOPO Viewer consists of four tab pages: TABLE, MAP, FILE and GFP. The TABLE tab page displays the prediction results for all the programs used.

Evola

Evola is a database of evolutionary annotation of human genes (22). It provides sequence alignments and phylogenetic trees of manually curated orthologous genes among human and 11 model organisms, Pan troglodytes (chimpanzee), Macaca sp. (macaque), Mus musculus (mouse), Rattus norvegicus (rat), Canis familiaris (dog), Bos taurus (cow), Monodelphis domestica (opossum), Gallus gallus (chicken), Danio rerio (zebra fish), Tetraodon nigroviridis (tetraodon) and Takifugu rubripes (fugu). Sequence alignments and phylogenetic trees of the orthologous genes and homologous genes are shown in Evola.

PPI view

The PPI view displays H-InvDB human PPI information at http://www.jbirc.aist.go.jp/hinv/ppi/. We collected PPI data from five databases; BIND, DIP, MINT, HPRD and IntAct, removed redundancies of the PPI data among the databases based on their sequence similarities and integrated them with the H-Invitational proteins.

Gene family/Group view

The Gene family/Group view provides human-curated annotation datasets for the selected gene families/groups at http://www.jbirc.aist.go.jp/hinv/ahg-db/geneFamilyIndex.jsp. For H-InvDB release 4.0, we provided detailed annotations for four selected gene families/groups: TCR, Ig, MHC and OR. Each page provides the list of genes, gene names, definitions and links for the appropriate H-InvDB views.

H-InvDB New Identifier

We defined and assigned a unique identifier for each annotation unit, transcript, protein or cluster (7,8). The identifier for H-Invitational transcript is ‘HIT’, prefix HIT plus nine digit numbers (e.g. HIT000000001) and for H-Invitational cluster is ‘HIX’, prefix HIX plus seven digit numbers (e.g. HIX0000001). In order to identify the modification in sequence or annotation of an H-Inv entry, a version is assigned to each ID and always stated with the ID. Additionally, we now provide a new identifier for each H-Invitational protein, ‘HIP’, prefix HIP with nine digit numbers (e.g. HIP000000001).

H-InvDB Data Availability

H-InvDB is freely available for both academic and commercial use and can be accessed online at http://www.h-invitational.jp/(or hinv.jp). Annotated data can also be downloaded in FASTA sequence files, the original-format flat files or XML files at HTTP and FTP servers. The mirror database is also available at http://hinvdb.ddbj.nig.ac.jp/. Minor updates are released every three months and major updates are released once a year.

LIST OF AUTHORS FOR THE GENOME INFORMATION INTEGRATION PROJECT AND H-INVITATIONAL 2 CONSORTIUM

Chisato Yamasaki^1,2, Katsuhiko Murakami^1,2, Yasuyuki Fujii³, Yoshiharu Sato^1,2, Erimi Harada^1,2, Jun-ichi Takeda^1,2, Takayuki Taniya^1,2, Ryuichi Sakate^1,2, Shingo Kikugawa^1,2, Makoto Shimada^1,2, Motohiko Tanino⁴, Kanako O. Koyanagi⁵, Roberto A. Barrero⁶, Craig Gough^1,2, Hong-Woo Chun^1,2, Takuya Habara¹, Hideki Hanaoka⁷, Yosuke Hayakawa^1,8, Phillip B. Hilton^1,2, Yayoi Kaneko⁹, Masako Kanno^1,2, Yoshihiro Kawahara^1,2, Toshiyuki Kawamura¹⁰, Akihiro Matsuya^1,11, Naoki Nagata¹², Kensaku Nishikata^1,13, Akiko Ogura Noda^1,2, Shin Nurimoto¹⁴, Naomi Saichi^1,2, Hiroaki Sakai¹⁵, Ryoko Sanbonmatsu^1,2, Rie Shiba^1,2, Mami Suzuki^1,2, Kazuhiko Takabayashi⁸, Aiko Takahashi^1,2, Takuro Tamura¹⁶, Masayuki Tanaka^1,2, Susumu Tanaka¹⁷, Fusano Todokoro^1,18, Kaori Yamaguchi¹, Naoyuki Yamamoto^1,19, Toshihisa Okido²⁰, Jun Mashima²⁰, Aki Hashizume²⁰, Lihua Jin²⁰, Kyung-Bum Lee²⁰, Yi-Chueh Lin²⁰, Asami Nozaki²⁰, Katsunaga Sakai²⁰, Masahito Tada²⁰, Satoru Miyazaki²¹, Takashi Makino²², Hajime Ohyanagi^20,23, Naoki Osato²⁰, Nobuhiko Tanaka²⁰, Yoshiyuki Suzuki²⁰, Kazuho Ikeo²⁰, Naruya Saitou²⁴, Hideaki Sugawara²⁰, Claire O’Donovan²⁵, Tamara Kulikova²⁵, Eleanor Whitfield²⁵, Brian Halligan²⁶, Mary Shimoyama²⁶, Simon Twigger²⁶, Kei Yura²⁷, Kouichi Kimura²⁸, Tomohiro Yasuda²⁸, Tetsuo Nishikawa^28,29, Yutaka Akiyama³⁰, Chie Motono³⁰, Yuri Mukai³⁰, Hideki Nagasaki^15,30, Makiko Suwa³⁰, Paul Horton³⁰, Reiko Kikuno³¹, Osamu Ohara³¹, Doron Lancet³¹, Eric Eveno^33,34, Esther Graudens^33,34, Sandrine Imbeaud^33,34,³⁵, Marie Anne Debily^33,34,³⁶, Yoshihide Hayashizaki^37,38, Clara Amid³⁹, Michael Han³⁹, Andreas Osanger³⁹, Toshinori Endo⁵, Michael A. Thomas⁴⁰, Mika Hirakawa⁴¹, Wojciech Makalowski⁴², Mitsuteru Nakao⁴³, Nam-Soon Kim⁴⁴, Hyang-Sook Yoo⁴⁴, Sandro J. De Souza⁴⁵, Maria de Fatima Bonaldo⁴⁶, Yoshihito Niimura⁴⁷, Vladimir Kuryshev⁴⁸, Ingo Schupp⁴⁸, Stefan Wiemann⁴⁸, Matthew Bellgard⁶, Masafumi Shionyu⁴⁹, Libin Jia⁵⁰, Danielle Thierry-Mieg⁵¹, Jean Thierry-Mieg⁵¹, Lukas Wagner⁵¹, Qinghua Zhang^34,52, Mitiko Go⁵³, Shinsei Minoshima⁵⁴, Masafumi Ohtsubo⁵⁴, Kousuke Hanada⁵⁵, Peter Tonellato⁵⁶, Takao Isogai²⁹, Ji Zhang^34,57, Boris Lenhard⁵⁸, Sangsoo Kim⁵⁹, Zhu Chen^34,60,⁶¹, Ursula Hinz⁶², Anne Estreicher⁶², Kenta Nakai⁶³, Izabela Makalowska⁶⁴, Winston Hide⁶⁵, Nicola Tiffin⁶⁵, Laurens Wilming⁶⁶, Ranajit Chakraborty⁶⁷, Marcelo Bento Soares⁶⁸, Maria Luisa Chiusano⁶⁹, Yutaka Suzuki⁷⁰, Charles Auffray^33,34, Yumi Yamaguchi-Kabata², Takeshi Itoh^2,15, Teruyoshi Hishiki², Satoshi Fukuchi²⁰, Ken Nishikawa²⁰, Sumio Sugano^2,70, Nobuo Nomura², Yoshio Tateno²⁰, Tadashi Imanishi^2,5,†, Takashi Gojobori^2,20