TruSight RNA Pan-Cancer Panel FAQs

Collapse All


Expand All

  • Input

  • The protocol is optimized for 10–100ng of human total RNA. Lower input amounts may result in low yield and reduced sensitivity. Use 10ng for high-quality universal human reference total RNA as input. 

    For FFPE RNA, the sample input amount is based on sample quality. Use the percentage of RNA fragments > 200 nt fragment distribution value (DV200) as a reliable determinant of FFPE RNA quality. 



    Input Requirement Per Reaction


    > 70%

    20 ng



    20–50 ng



    50–100 ng

    Too degraded

    < 30%

    Not recommended

    For more information, see the Evaluating RNA Quality from FFPE Samples tech note.

    For successful library preparation, use an RNA isolation method that includes a reverse-crosslinking step and DNase1 treatment, such as the QIAGEN RNeasy FFPE kit or the QIAGEN AllPrep DNA/RNA FFPE kit.

    For samples that border quality classifications, use the higher end of the input recommendation.

    Success is not guaranteed with poor quality libraries. For libraries with DV200 < 30%, input of 100 ng or greater is recommended. 

    Use NanoDrop to quantify the RNA concentration. If NanoDrop is not available, use a fluorometric method.

  • Analysis

  • The BaseSpace RNA-Seq Alignment App performs alignment and analyzes for gene fusions with STAR or TopHat. STAR is recommended for optimal fusion calling.

    The BaseSpace TopHat Alignment App can also be used for alignment and fusion calling. The BaseSpace Cufflinks Assembly & DE App can be used to run differential expression analysis.

    Third-party analysis tools are also available.

    The main difference between the apps is that the RNA-Seq Alignment App is specifically designed for optimal fusion calling using STAR.

    You can try both analysis methods and compare results.

    Run the RNA-Seq workflow (FASTQ only) on the MiSeq and stream the data to BaseSpace. The BaseSpace RNA-Seq Alignment App analyzes data from the TruSight RNA Pan-Cancer Panel, providing a simple results summary that includes a fusion table, variant table, and gene expression table.

    You can also use your own pipeline for analysis.

    The STAR aligner in the RNA-Seq Alignment App as been optimized for fusion calling and is recommended.

    You can also use the TopHat Alignment App. Running both analysis options and comparing the data can be useful.

    Use the Homo sapiens (PAR-masked)/hg19 (RefSeq) or Homo sapiens (PAR-masked)/hg 19 (Gencode) reference genome. Because of different annotations within these genomes, results can vary by reference.

    In BaseSpace, click your Projects folder. Select Analysis and select the app session name that you saved your analysis under.

    Yes, there are example data sets in BaseSpace Public Data.

    The expected range of percent duplicates is ≤ 25% for controls (UHR) at 0.5M subsampled reads.

    The percent of reads passing filter (PF) aligned to human ribosomal RNA is typically ≤ 8%.

    The BaseSpace RNA-Seq Alignment App with the STAR aligner may call a fusion if there are at least three unique reads that meet all the quality metrics, including the following threshold filters.

    • Split Reads + Paired Reads ≥ 3
    • Alt/Ref Reads ≥ 0.01
    • Fusion Contig Align Length (bp) > 16
    • Break-end Homology (bp) ≤ 10
    • Alternative Local Contig Align Fraction < 0.8
    • Coverage after fusion (bp) ≥ 100

    However, a high number of nonfusion supporting reads (ie, wild type transcript) in that region would be expected to cause noise that can affect fusion calling.

    A paired read is a fusion where one read aligns to the left gene and the other read aligns to the right gene.

    A split read is a fusion where one of the reads spans the fusion junction.

    The best way to confirm that an identified fusion is 'real' is to use an orthogonal approach. Additionally, assess the quality score and the chromosomal locations of the fusions to help indicate confidence.

    To confirm fusion calls, perform additional investigation or independent validation. Fusion transcripts between nearby genes on the same chromosome and strand can be a result of read-through transcription rather than genomic translocation. To reduce these calls, the fusion software is optimized to filter read-through transcripts, but can result in filtering of biologically relevant fusions between adjacent genes (eg, STIL-TAL1). Also, fusions called between genes of high homology, such as a gene and its pseudogene, can be artifacts of multiple alignment instead of genomic rearrangements. Illumina reccommends that the quality score and chromosomal location be assessed and confirmed using independent molecular biology approaches.

    Several factors can cause a fusion to not get called:
    - Low expression levels of the fusion gene. More sequencing read depth or lower fusion score threshold may be required.
    - Low quality of the sample. More sample input RNA may be required.
    - Close proximity of 2 genes in the same orientation, on the same chromosome (eg, STIL-TAL1). Reducing the default breakpoint distance thresholds in the Local Run Manager module can help identify these fusions, but may also display more false positive calls.
    - Differences in bioinformatics algorithms used.
    - Differences in reference genomes. GENCODE has higher genomic coverage than RefSeq and fusions between exons that are not annotated are not called in RefSeq. For example, an EML4 transcript has an exon that is annotated in GENCODE but not RefSeq, meaning fusions at that exon would not be called when using the RefSeq reference.

    Chromosomal translocations resulting in overexpression or deletion of a transcript can be reflected in gene expression levels but would not create a fusion gene. The Local Run Manager RNA Fusion module is not designed for detection of gene expression changes. To detect these changes, the RNA-Seq Alignment App in BaseSpace Sequence Hub is recommended. In addition, it is recommended to confirm these findings in DNA.

    Yes, use the RNA-Seq Differential Expression BaseSpace Sequence Hub app.

    No, only one of the gene fusion partners needs to be detected. The enrichment approach allows you to pull down the target and the partner fusion gene with it. 

  • Results

  • Sensitivity of fusion detection varies depending on the expression level of the fusion transcript in the sample. Illumina performed titration experiments with synthetic fusion constructs. Spike-in fusions were detected as low as concentrations equivalent to ~13 fusion copies/cell (1 E-7 pM) (based on 500 cells/10ng input).

    The RNA-Seq Alignment App requires at least three unique fusion-supporting reads to call a fusion positively (with some additional quality and read metrics). However, a high number of nonfusion supporting reads in that region is expected to cause noise that can affect fusion calling.