Limits...
Identification of novel fusion genes in lung cancer using breakpoint assembly of transcriptome sequencing data.

Fernandez-Cuesta L, Sun R, Menon R, George J, Lorenz S, Meza-Zepeda LA, Peifer M, Plenker D, Heuckmann JM, Leenders F, Zander T, Dahmen I, Koker M, Schöttle J, Ullrich RT, Altmüller J, Becker C, Nürnberg P, Seidel H, Böhm D, Göke F, Ansén S, Russell PA, Wright GM, Wainer Z, Solomon B, Petersen I, Clement JH, Sänger J, Brustugun OT, Helland Å, Solberg S, Lund-Iversen M, Buettner R, Wolf J, Brambilla E, Vingron M, Perner S, Haas SA, Thomas RK - Genome Biol. (2015)

Bottom Line: Genomic translocation events frequently underlie cancer development through generation of gene fusions with oncogenic properties.Identification of such fusion transcripts by transcriptome sequencing might help to discover new potential therapeutic targets.We apply TRUP to RNA-seq data of different tumor types, and find it to be more sensitive than alternative tools in detecting chimeric transcripts, such as secondary rearrangements in EML4-ALK-positive lung tumors, or recurrent inactivating rearrangements affecting RASSF8.

View Article: PubMed Central - PubMed

ABSTRACT
Genomic translocation events frequently underlie cancer development through generation of gene fusions with oncogenic properties. Identification of such fusion transcripts by transcriptome sequencing might help to discover new potential therapeutic targets. We developed TRUP (Tumor-specimen suited RNA-seq Unified Pipeline) (https://github.com/ruping/TRUP), a computational approach that combines split-read and read-pair analysis with de novo assembly for the identification of chimeric transcripts in cancer specimens. We apply TRUP to RNA-seq data of different tumor types, and find it to be more sensitive than alternative tools in detecting chimeric transcripts, such as secondary rearrangements in EML4-ALK-positive lung tumors, or recurrent inactivating rearrangements affecting RASSF8.

Show MeSH

Related in: MedlinePlus

Comparison of TopHat-Fusion with TRUP analysis tools. The fusion candidates found by TRUP and TopHat-Fusion on sample S00054, plotted by the number of non-redundant spanning reads reported by the two tools. The horizontal and vertical dashed lines indicate the threshold of three spanning reads for calling a chimeric transcript. The diagonal dashed line is plotted to show that TRUP usually reports more reads spanning a fusion point. The axes are in log2 scale. The unique calls to each algorithm are jittered to avoid over-plotting.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4300615&req=5

Fig4: Comparison of TopHat-Fusion with TRUP analysis tools. The fusion candidates found by TRUP and TopHat-Fusion on sample S00054, plotted by the number of non-redundant spanning reads reported by the two tools. The horizontal and vertical dashed lines indicate the threshold of three spanning reads for calling a chimeric transcript. The diagonal dashed line is plotted to show that TRUP usually reports more reads spanning a fusion point. The axes are in log2 scale. The unique calls to each algorithm are jittered to avoid over-plotting.

Mentions: We compared TopHat-Fusion and TRUP in more detail since they showed the best performance (the harmonic mean of recall and precision for TRUP is 0.62 and for TopHat-Fusion is 0.49, highest among all the tools) with a total of 20 and 40 predicted fusion events, respectively (Additional files 8 and 9). For candidates with very low coverage, disagreements between the two tools were observed, indicating higher uncertainty for calling fusion transcripts with low expression. After manually checking the calls unique to TRUP, we only found a single candidate that might be considered as a false positive. This candidate exhibited a breakpoint located in a repetitive region and also showed a low spanning score, which summarizes the confidence of supporting evidence of the spanning reads (Materials and methods). In order to avoid using suboptimal parameter setting for TopHat-Fusion, we alternatively used the default settings and adjusted the TRUP parameters accordingly. We therefore increased the threshold for fusion calling as follows: presence of at least three reads spanning the fusion point and two encompassing mate pairs. TopHat-Fusion now detected eight fusion events all of which were included in the 25 candidates found by TRUP (Figure 4). Both tools successfully recovered the EML4-ALK fusion as well as one of the secondary fusions, SNAP29-CELSR1. However, the fusion event PIGF-CHMP3 was only reported by TRUP. TopHat-Fusion failed to call this true positive because the number of spanning reads was limited to two. By contrast, TRUP detected nine fusion-spanning reads. We found that TRUP usually reports more non-redundant spanning reads than TopHat-Fusion, indicating a higher sensitivity in identifying reads showing chimeric patterns. Judging from the results of the analysis of sample H3122, TRUP performs well even on paired-end data with short insert sizes (here 70 bp) where mate pairs overlap.Figure 4


Identification of novel fusion genes in lung cancer using breakpoint assembly of transcriptome sequencing data.

Fernandez-Cuesta L, Sun R, Menon R, George J, Lorenz S, Meza-Zepeda LA, Peifer M, Plenker D, Heuckmann JM, Leenders F, Zander T, Dahmen I, Koker M, Schöttle J, Ullrich RT, Altmüller J, Becker C, Nürnberg P, Seidel H, Böhm D, Göke F, Ansén S, Russell PA, Wright GM, Wainer Z, Solomon B, Petersen I, Clement JH, Sänger J, Brustugun OT, Helland Å, Solberg S, Lund-Iversen M, Buettner R, Wolf J, Brambilla E, Vingron M, Perner S, Haas SA, Thomas RK - Genome Biol. (2015)

Comparison of TopHat-Fusion with TRUP analysis tools. The fusion candidates found by TRUP and TopHat-Fusion on sample S00054, plotted by the number of non-redundant spanning reads reported by the two tools. The horizontal and vertical dashed lines indicate the threshold of three spanning reads for calling a chimeric transcript. The diagonal dashed line is plotted to show that TRUP usually reports more reads spanning a fusion point. The axes are in log2 scale. The unique calls to each algorithm are jittered to avoid over-plotting.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4300615&req=5

Fig4: Comparison of TopHat-Fusion with TRUP analysis tools. The fusion candidates found by TRUP and TopHat-Fusion on sample S00054, plotted by the number of non-redundant spanning reads reported by the two tools. The horizontal and vertical dashed lines indicate the threshold of three spanning reads for calling a chimeric transcript. The diagonal dashed line is plotted to show that TRUP usually reports more reads spanning a fusion point. The axes are in log2 scale. The unique calls to each algorithm are jittered to avoid over-plotting.
Mentions: We compared TopHat-Fusion and TRUP in more detail since they showed the best performance (the harmonic mean of recall and precision for TRUP is 0.62 and for TopHat-Fusion is 0.49, highest among all the tools) with a total of 20 and 40 predicted fusion events, respectively (Additional files 8 and 9). For candidates with very low coverage, disagreements between the two tools were observed, indicating higher uncertainty for calling fusion transcripts with low expression. After manually checking the calls unique to TRUP, we only found a single candidate that might be considered as a false positive. This candidate exhibited a breakpoint located in a repetitive region and also showed a low spanning score, which summarizes the confidence of supporting evidence of the spanning reads (Materials and methods). In order to avoid using suboptimal parameter setting for TopHat-Fusion, we alternatively used the default settings and adjusted the TRUP parameters accordingly. We therefore increased the threshold for fusion calling as follows: presence of at least three reads spanning the fusion point and two encompassing mate pairs. TopHat-Fusion now detected eight fusion events all of which were included in the 25 candidates found by TRUP (Figure 4). Both tools successfully recovered the EML4-ALK fusion as well as one of the secondary fusions, SNAP29-CELSR1. However, the fusion event PIGF-CHMP3 was only reported by TRUP. TopHat-Fusion failed to call this true positive because the number of spanning reads was limited to two. By contrast, TRUP detected nine fusion-spanning reads. We found that TRUP usually reports more non-redundant spanning reads than TopHat-Fusion, indicating a higher sensitivity in identifying reads showing chimeric patterns. Judging from the results of the analysis of sample H3122, TRUP performs well even on paired-end data with short insert sizes (here 70 bp) where mate pairs overlap.Figure 4

Bottom Line: Genomic translocation events frequently underlie cancer development through generation of gene fusions with oncogenic properties.Identification of such fusion transcripts by transcriptome sequencing might help to discover new potential therapeutic targets.We apply TRUP to RNA-seq data of different tumor types, and find it to be more sensitive than alternative tools in detecting chimeric transcripts, such as secondary rearrangements in EML4-ALK-positive lung tumors, or recurrent inactivating rearrangements affecting RASSF8.

View Article: PubMed Central - PubMed

ABSTRACT
Genomic translocation events frequently underlie cancer development through generation of gene fusions with oncogenic properties. Identification of such fusion transcripts by transcriptome sequencing might help to discover new potential therapeutic targets. We developed TRUP (Tumor-specimen suited RNA-seq Unified Pipeline) (https://github.com/ruping/TRUP), a computational approach that combines split-read and read-pair analysis with de novo assembly for the identification of chimeric transcripts in cancer specimens. We apply TRUP to RNA-seq data of different tumor types, and find it to be more sensitive than alternative tools in detecting chimeric transcripts, such as secondary rearrangements in EML4-ALK-positive lung tumors, or recurrent inactivating rearrangements affecting RASSF8.

Show MeSH
Related in: MedlinePlus