© 2026. All rights reserved.
© 2026. All rights reserved.
Florence Rufflé, Jérôme Audoux, Anthony Boureux, Sacha Beaumeunier, Jean-Baptiste Gaillard, Elias Bou Samra, Andre Megarbane, Bruno Cassinat, Christine Chomienne, Ronnie Alves, Sébastien Riquier, Nicolas Gilbert, Jean-Marc Lemaitre, Delphine Bacq-Daian, Anne Laure Bougé, Nicolas Philippe and Thérèse Commes.
see the article at F1000Research 2017, 6(ISCB Comm J):1302 - doi: 10.12688/f1000research.11352.1.
from Charles Gawad, Department of Computational Biology, St. Jude Children’s Research Hospital, Memphis, TN, USA
In this report, Ruffle et al. present a new approach for identifying novel chimeric transcripts using CracTools. They classify the transcripts into four categories: 1) different chromosomes (interchromosomal translocations), 2) co-linear but from different genes (putative transcriptional read through or complex intrachromosomal structural variation), 3) exons from same chromosome but are not in the expected order based on the reference (tandem duplication, complex intrachromosomal rearrangement), or 4) exons on same chromosome but different strands. The authors then go on to validate a subset of putative new chimeric transcripts using RT-PCR and FISH. These types of studies have been previously performed. One new aspect of this study is the stranded library preparation which allows for the identification of the class 4 chimeras. In addition, they tried to minimize the bias in their analysis pipeline by not relying on a reference genome, which enabled the discovery of new transcripts. Overall, their approach provides a validated new strategy for identifying novel transcripts in RNA-seq data.
We thank the referee for his careful reading of our manuscript and for his remarks which led us to clarify the protocol used in this work.
- The authors do not discuss circular RNA, which are likely to make up a large portion of their class 3 chimeras as found in many recent studies. The low numbers of class 3 chimeras also raises concerns about the sensitivity of the approach, as most recent studies have found thousands of circular RNA isoforms per sample. The total RNA underwent RT with random primers, which would retain circular RNA. It is not clear if there was a polyA-selection step after that point or if the total RNA was ribosomal depleted. If it was the former, the circular transcripts would not be present.
Response: The remark is relevant and we agree with the referee that it was not clearly stated that we performed polyA selection for the RNAseq experiment. This point is only described in online data availability (https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-5767/) and doesn’t appear explicitly in mat&methods section. We added the following comment on page 7: “The RNAseq was performed using polyA-selection with the TruSeq RNA Lib-Prep Kit (Illumina) adjusted with GATC specific procedure for strand specificity”. We also agree with the referee that class3 fusion transcripts could arise from circular RNA (circRNA) when performing ribosomal depleted RNA-seq. However, as our polyA+ RNA-seq study doesn’t enable the identification of this RNA subtype, we do not discuss about circRNA in the manuscript. Concerning class3 chimeric RNA, our pipeline detects a great number of candidates before applying specific filters. This category, as well as the class2, is the most represented in all datasets we analyzed. As described in mat&methods section, we used stringent criteria to minimize bias and greatly reduced the number of candidates. For example, from a typical AML stranded and paired- end RNA-seq experiment (50 Millions reads; 100pb length), starting from the raw data (CRAC and CracTools process) we extracted 1406 chRNAs including 35% of class3chRNA reduced by 16 fold with the filtering process (26 class3chRNA).
- If the RNA was not polyA-selected, the authors should specifically discuss the PML-RARA circular transcripts recently discovered in acute promyelocytic leukemia. I am not aware of any independent validation of that work.
Response: As we choose polyA RNA-seq, we can only characterize new linear fusion transcripts. From ribo-depleted RNAseq data analysis, the characterization of circRNA arising from fusion gene requires the identification of linear junction (lin-J) and circular junction (circ-J) described as spliced and back spliced junction respectively by Guarnerio et al (Oncogenic Role of Fusion-circRNAs Derived from Cancer-Associated Chromosomal Translocations, 2016; Cell 165 (2): 289-302). The linear and circular fusion transcripts share the same linear junction. The new (3-12) PML-RARA junction we discover corresponds necessarily to a linear junction because of its fusion junction sequence (see also the joined figure). Moreover, in our polyA+ dataset it surely corresponds to a linear transcript. It is most probably an alternative splicing product transcribed from the PML-RARA bcr3 translocation since it coexists with the well-known (3-9) PML-RARA bcr3 transcript. However we could not exclude in the biological sample the presence of circRNA emerging from the PML-RARA fusion gene. However polyA RNAseq protocol doesn’t enable to reveal them. To answer to the referee remark, we performed complementary experiments and analyzed the Guarnerio dataset (BioProjectID: PRJNA315254) with our pipeline to detect fusion junctions (lin-J and Circ-J) specific to circRNA arising from the PML-RARA translocated genes. We found 3 PML-RARA and 1 RARA-PML linear fusion junctions but we did not detect the circRNA ones. We performed a tag search approach specific to the PML-RARA circ-junction (F-circ1) described in Fig S1B supplemental information of Guarnerio et al manuscript. We did not find it in their fastQ files. To conclude we were unable to detect PML-RARA cirRNA in Guarnerio J dataset but we confirmed the data recently described by You, X. and Conrad, T. OF with the specific cirRNA Acfs pipeline (Acfs: accurate circRNA identification and quantification from RNA-Seq data, Scientific Report 6,38820; doi: 10.1038/srep38820; 2016). The circRNA search for low abundance will certainly require more deepness in RNAseq.
A- PML and RARA genes schematic view with alternating exons and introns. Filled
rectangles represent exons and bcr1, bcr2, bcr3 the potential PML break chromosomic
region described in PML-RARA translocation.
B- PML-RARA and RARA-PML fusion genes resulting from bcr3 breakpoint. ChRNAs
junctions arising from these constructs are indicated (Lin-J for linear junction
and Circ-J for circular junction). The new 3-12 linear junction relies 3’ end of PML
exon 3 with 5’ start of RARA exon 12. The 3-12 circular
junction, if exists, would join 3’ end of RARA exon 12 with 5’ start of PML exon 3.
C- ChRNAs read sequences with the corresponding linear junctions.
- The authors should read the manuscript closely for typos. For example in the AML samples and cells lines section there are commas where there should be periods, in the FISH methods section there are degree signs instead of percent, and two paragraphs before the discussion there is MDR instead of MRD.
Response: We read carefully the manuscript and corrected the typos.
- The authors should include the kits used for ribosomal RNA-depletion/polyA-selection, as well as stranded library preparation.
Response: We added the following comment on page 4: “The RNAseq was performed using polyA-selection with the TruSeq RNA Lib-Prep Kit (Illumina, San Diego, CA) adjusted with GATC specific procedure for strand specificity”.
- What mechanisms do the authors have in mind for a structural variant and/or alternative splicing event that would result in class 4 chimeras, as they would not occur as a result of transcriptional read through of the same strand?
Response: Two mechanisms would result in class 4 chimeras. The first one could involve chromosomic duplication and inversion as described in Newman et al (Next-generation sequencing of duplication CNVs reveals that most are tandem and some create fusion genes at breakpoints, Am J Hum Genet. 2015, Feb 5; 96(2):208-20. doi:10.1016/j.ajhg. 2014.12.017. Epub 2015 Jan 29). The second one could involve splicing events as described by Gingeras et al (Implications of chimaeric non-co-linear transcripts, Nature. 2009 Sep 10; 461(7261):206-11doi: 10.1038/nature08452.)
- The resolution is too low for Figure 5A.
Response: We changed the resolution of the figure 5A, increasing the font size.