Cancer Res: A new method for accurate and comprehensive detection of cancer gene fusion events based on long-read sequencing: FusionSeeker

Recently, a research team at the University of Alabama at Birmingham proposed a gene fusion detection tool, FusionSeeker, which can accurately and comprehensively identify gene fusion events in long-reading cancer transcriptome sequencing data and reconstruct accurate fusion transcripts
from original reads.
FusionSeeker detects gene fusion tunings based on reads number and filters them to remove noisy signals
due to sequencing errors and reads alignment errors.
FusionSeeker has been validated to recognize gene fusions in exon and intron regions, allowing for comprehensive characterization
of gene fusions in the cancer transcriptome.

The findings were published in Cancer Research in an article titled "Gene fusion detection and characterization in long-read cancer transcriptome sequencing data with FusionSeeker.
"
。

The research team first benchmarked the accuracy of the FusionSeeker gene fusion detection on a simulated dataset, randomly generating a total of 150 gene fusion transcripts and assigning them to different expression levels
.
PacBio Iso-Seq-like and Nanopore-like reads with PBSIM and Badread (v0.
2.
0 simulate, then align to the reference genome
.
FusionSeeker and long-chain gene fusion detection tools JAFFAL and LongGF are used to detect gene fusion from simulated reads
。 The team repeated the simulation three times, and FusionSeeker identified more
true positive events.

At the same time, FusionSeeker's high recall is mainly due to its ability to detect gene fusions in intron regions, in which FusionSeeker identified 94.
67% of intron events, JAFFAL and LongGF reported 14.
67% and 54.
67%,
respectively.
Overall, the above three fusion detection tools can achieve higher recall when detecting high and medium expression levels due to fusion
.
About 67 percent of the gene fusions lost by FusionSeeker came from the low-expression group due to low reads coverage
.

Figure 1: FusionSeeker inspection process

To generate highly accurate transcript sequences, FusionSeeker performs partial order alignment using reads containing fusions, calculating consistent sequences
for each gene fusion event.
In the simulated dataset, FusionSeeker reconstructed full-length transcripts of more than 99.
5% of fusion events, using Iso-Seq and Nanopore reads The average sequence recognition rates were 99.
87% and 99.
14%,
respectively.
It is shown that FusionSeeker can accurately identify gene fusions and report full-length fusion transcript sequences
in simulated datasets.

Next, the research team evaluated the false discovery rates
of the three tools using the non-cancer dataset of the Human Genome Structural Variation Consortium (HGSVC).
Of all 12 non-cancer samples, FusionSeeker reported the lowest number of gene fusions, indicating that FusionSeeker had the lowest rate of false findings among the
three gene fusion detection tools.
The research team also applied FusionSeeker to the analysis of samples from an acute myeloid leukemia (AML) patient to demonstrate its clinical utility
.
Ultimately, FusionSeeker identified a pre-validated gene fusion between RUNX1 and RUNX1T1 and reported another 7 in the patient's sample A credible gene fusion event
.

Finally, the team evaluated transcript sequences
generated by FusionSeeker.
In the Iso-Seq and Nanopore datasets of the MCF-7 cell line compared to the original reads, FusionSeeker The consistency of transcript sequences with reference gene sequences was significantly improved (Figure 2).

In the Iso-Seq dataset of SKBR3 and the Nanopore dataset of the HCT-116 cell line The transcript sequences reported by FusionSeeker were more accurate
than the original reads.

Figure 2: FusionSeeker transcript sequence versus reference gene sequence

In summary, FusionSeeker can detect gene fusions
in exon and intron regions.
Based on simulations and data from three cancer cell lines, the study demonstrated that FusionSeeker outperforms existing methods
in detecting gene fusion events.
In addition, the research team verified many gene fusion events
that were only detected by FusionSeeker through orthogonal and experimental validation.
These new gene fusions may have important implications for tumor genesis and progression and warrant further study
.
Overall, FusionSeeker will enable users to accurately discover gene fusions using long-read sequencing data, facilitate downstream functional analysis, and improve cancer diagnosis and treatment
.

References:

Yu Chen, Yiqing Wang, Weisheng Chen,et al.
Gene fusion detection and characterization in long-read cancer transcriptome sequencing data with FusionSeeker.
Cancer Res CAN-22-1628.
2022.
https://doi.
org/10.
1158/0008-5472.
CAN-22-1628.