This track is part of the Fusion Genes collection. It shows fusion genes from FusionGDB, a database of human fusion genes and their predicted effect on the encoded protein. The track contains about 53,000 distinct breakpoint pairs of 48,000 fusion genes.
The breakpoints come from three data sets:
Each fusion is shown as two features, one for each partner gene. A partner is drawn as the part of its transcript that is kept in the fusion transcript: for the 5' partner from the start of the transcript to the breakpoint, for the 3' partner from the breakpoint to the end of the transcript. The thick part is the coding sequence that is kept. The same fusion gene can have several breakpoint pairs, and every pair is a separate pair of features. Each feature has an ID made of the fusion name, the number of the breakpoint pair and the partner, e.g. TMPRSS2::ERG.3.5p, shown on the details page.
The breakpoint end of each feature is marked with a black bar and an orange arrow that points in the direction of transcription: in the 5' partner towards the breakpoint, in the 3' partner away from it. For EML4::ALK, for example, the arrow on EML4 points to the breakpoint and the one on ALK points from the breakpoint into ALK.
Hover over a feature to see its partner gene, reading frame, number of TCGA samples and sources. The mouseover and the details page link to the partner gene, which is then highlighted. The details page links to the FusionGDB page of the fusion gene, if FusionGDB has one, and the TCGA samples link to the patient pages on cBioPortal. When zoomed out so far that a feature is only a few pixels wide, a small flag next to the black bar shows on which side of the breakpoint the feature is.
Features are colored by the effect of the fusion on the reading frame. FusionGDB annotated each breakpoint pair for all pairs of GENCODE V19 transcripts of the two genes. If at least one transcript pair is in-frame, the feature is shown as in-frame.
| In-frame: both breakpoints are in coding exons and the fusion transcript keeps the reading frame of the 3' gene | |
| Frame-shift: both breakpoints are in coding exons, but the fusion shifts the reading frame of the 3' gene | |
| Outside CDS: at least one breakpoint is in a UTR or intron in all transcript pairs, or FusionGDB has no reading frame annotation for this breakpoint pair |
A breakpoint outside the coding sequence does not mean that the fusion makes no protein. In TMPRSS2::ERG, for example, the 5' UTR of TMPRSS2 is joined to ERG, and ERG protein is made from its own start codon under the control of the TMPRSS2 promoter. The codon frames shown when codon coloring is switched on are those of the original transcript, which is the frame of the fusion protein only for in-frame fusions.
FusionGDB gives the strand of the fusion sequence, not of the genes. For many ChiTaRS fusions, which come from GenBank ESTs and mRNAs, the GenBank sequence is the reverse complement of the transcript. In these cases both genes are on the other strand than given in FusionGDB, the 3' partner comes first in the transcript, and we draw the fusion with the roles of the two genes swapped. If only one of the two genes is on the other strand, the fusion is drawn as given by FusionGDB. Both cases are noted in the mouseover and on the details page.
The track can be filtered by reading frame, source, TCGA cancer type and number of TCGA samples. Fusion names such as TMPRSS2::ERG, TMPRSS2-ERG or TMPRSS2_ERG can be typed into the position box.
FusionGDB collected fusion breakpoints from ChiTaRS 3.1 (17,860 fusion genes), TumorFusions (15,854 fusion genes in 5,517 TCGA samples) and the TCGA PanCanAtlas fusion calls of Gao et al. (23,944 fusion genes in 6,071 samples), all on hg19. For every breakpoint pair, FusionGDB located both breakpoints in all GENCODE V19 transcripts of the two genes. A transcript pair was called in-frame if both breakpoints are in coding exons and the length of the fusion transcript from the start codon of the 5' gene to the 3' gene's stop codon is a multiple of three, and frame-shift if both are in coding exons but it is not. Other combinations were labeled by the breakpoint location, e.g. 5UTR-3CDS. See Kim and Zhou 2019 for details.
We downloaded the files TCGA_ChiTaRS_combined_fusion_information_on_hg19.txt and TCGA_ChiTaRS_combined_fusion_ORF_analyzed_gencode_h19v19.txt from the FusionGDB download page. The first file has 65,780 rows, one per fusion, sample and source. We merged rows with identical breakpoints in both genes into 52,994 breakpoint pairs. The breakpoint positions in the file are 1-based and are the last kept base of the 5' partner and the first kept base of the 3' partner. For the TCGA data, about 95% of them are the last base of an exon of the 5' gene or the first base of an exon of the 3' gene in GENCODE V19. In 205 breakpoint pairs from ChiTaRS, the chromosome of one partner gene is unknown in FusionGDB, and only the other partner is shown.
For each breakpoint pair, we took the GENCODE V19 transcript pair with the best reading frame (in-frame, then frame-shift, then any other) and, among those, the transcripts tagged appris_principal, then CCDS, then basic, then the one with the longest coding sequence. For breakpoint pairs without frame annotation, we took the best GENCODE V19 transcript of the gene by the same rules, among the transcripts that contain the breakpoint. Each transcript was cut at the breakpoint. In FusionGDB, 183 gene symbols from ChiTaRS had been converted to dates by a spreadsheet program, e.g. 9-Sep for SEPT9 or 1-Sep-04 for SEPT14. We changed them back to the GENCODE V19 symbol (SEPT, MARCH, MARC or DEC genes) whose transcript contains the breakpoint. Of the 105,783 partner features on hg19, 90,986 are drawn from a FusionGDB transcript pair and 8,334 from a GENCODE transcript of the gene. For 6,463, no GENCODE V19 transcript of the gene contains the breakpoint, and only the breakpoint is shown as a 1 bp feature. In 3,451 fusion events both genes are on the other strand than the fusion sequence and the roles were swapped, and in 8,059 only one gene is. FusionGDB's in-frame annotation is not always the same as a strict reading frame check: for about a quarter of the in-frame breakpoint pairs, the number of coding bases kept from the 5' partner does not match the codon frame at the breakpoint of the 3' partner.
For hg38, the features were converted with liftOver; 84 features could not be converted and are not shown. Mitochondrial breakpoints are on the rCRS sequence, called chrMT on hg19 and chrM on hg38, and were not converted. The commands are in the makeDoc file and the scripts are in our source repository.
The FusionGDB data is free for academic research. Its terms of use do not allow redistribution, so this track cannot be exported with the Table Browser. The original data can be downloaded from the FusionGDB download page. Commercial use requires a license from FusionGDB.
The download page has tab-separated text files, all with hg19 coordinates. The main files are:
Thanks to Pora Kim and the FusionGDB team at the University of Texas Health Science Center at Houston for making the data available.
Kim P, Zhou X. FusionGDB: fusion gene annotation DataBase. Nucleic Acids Res. 2019 Jan 8;47(D1):D994-D1004. PMID: 30407583; PMC: PMC6323909
Kim P, Tan H, Liu J, Lee H, Jung H, Kumar H, Zhou X. FusionGDB 2.0: fusion gene annotation updates aided by deep learning. Nucleic Acids Res. 2022 Jan 7;50(D1):D1221-D1230. PMID: 34755868; PMC: PMC8728198
Hu X, Wang Q, Tang M, Barthel F, Amin S, Yoshihara K, Lang FM, Martinez-Ledesma E, Lee SH, Zheng S et al. TumorFusions: an integrative resource for cancer-associated transcript fusions. Nucleic Acids Res. 2018 Jan 4;46(D1):D1144-D1149. PMID: 29099951; PMC: PMC5753333
Gao Q, Liang WW, Foltz SM, Mutharasu G, Jayasinghe RG, Cao S, Liao WW, Reynolds SM, Wyczalkowski MA, Yao L et al. Driver Fusions and Their Implications in the Development and Treatment of Human Cancers. Cell Rep. 2018 Apr 3;23(1):227-238.e3. PMID: 29617662; PMC: PMC5916809
Gorohovski A, Tagore S, Palande V, Malka A, Raviv-Shay D, Frenkel-Morgenstern M. ChiTaRS-3.1-the enhanced chimeric transcripts and RNA-seq database matched with protein-protein interactions. Nucleic Acids Res. 2017 Jan 4;45(D1):D790-D795. PMID: 27899596; PMC: PMC5210585