Visualizing RNA-seq alignments with fusions with IGV

Step 1: Set Up Split-Screen View in IGV

Fusions involve two distinct genomic loci (e.g., TMPRSS2 on Chr 21 and ERG on Chr 21, or two different chromosomes). 1. Go to the IGV search bar at the top. 2. Enter both gene locations separated by a space (e.g., TMPRSS2 ERG or chr21:41465000 chr21:38400000). 3. Press Enter. IGV will create a split panel view showing Gene 1 on the left and Gene 2 on the right.

Step 2: Configure IGV Display Settings

Right-click on the BAM track coverage/alignment panel and make sure the following settings are enabled: - Color alignments by → SA tag (already done) or pair orientation / insert size (crucial for pair-end discordant reads). - View as pairs: Enable this so paired-end reads are visually connected by a thin line. - Show soft-clipped bases: Enable this (Right-click track → Preferences → Alignments → check Show soft-clipped bases).

Step 3: Identify & Count the Evidence

A. Split Reads (Direct Junction Overlap)

These are single sequencing reads that span across the fusion breakpoint. STAR aligns one part to Gene 1 and soft-clips the remaining sequence, which maps to Gene 2 as a supplementary alignment. - Visual Signature: - The read will display colored soft-clipped bases at the exact junction coordinate reported by Arriba.Right-click a split read and choose “View mate in split screen” or “Go to mate / supplementary alignment”. The other portion of the read will highlight at the exact junction coordinate of Gene 2. - How to Count: Count every unique read ID that has a soft-clipped tail precisely terminating at the breakpoint position. -

B. Discordant Mates (Spanning Pairs)

These are paired-end reads where Read 1 maps entirely within Gene 1, and Read 2 maps entirely within Gene 2. - Visual Signature: - When set to Color alignments by insert size/pair orientation, these pairs will turn colored (usually teal, blue, or red instead of standard gray) because their insert size or chromosome mapping is abnormal. - Dragging across the split screen will show thin lines connecting Read 1 in Gene 1 to Read 2 in Gene 2 across the panel break.

How to Count: Count the number of uniquely colored pair connections bridging Gene 1 and Gene 2.

Step 4: Validate Against Arriba/STAR-fusion Outputs

Compare your visual count to the metrics in Arriba or STAR-Fusion’s output file (fusions.tsv): - Check Breakpoints: Verify that the soft-clip boundaries match the breakpoint1 and breakpoint2 genomic coordinates in fusions.tsv. - Compare Read Counts:split_reads1 / split_reads2: Should equal the number of split reads you count anchoring at Gene 1 and Gene 2 respectively. - discordant_mates: Should equal the count of uniquely mapped paired-end fragments bridging the two genes.

Step 5: Rule Out Artifacts (False Positive Filters)

To ensure the fusion is real and not an alignment artifact:Mapping Quality (MAPQ):

Hover over the supporting reads. Make sure the MAPQ is high (>20–30).
Soft-clipped reads with MAPQ 0 are usually caused by repetitive elements or homologous sequences.
Anchor Length: Ensure the split-read segments mapped to both sides of the junction have a reasonable anchor length (>15–20 bp), rather than just a few soft-clipped nucleotides.
Strand Orientation: Check if the reads align continuously with respect to the strand logic (e.g., sense-to-sense for exon-exon fusions).
Unmapped Mates: Beware if the region contains a high density of multi-mapping reads or unaligned sequences (colored purple/white mismatch bars), which points toward library preparation or mapping artifacts.

References

https://okko73313.blogspot.com/2014/10/visualizing-rna-seq-alignments-with.html
https://help.connected.illumina.com/dragen/dragen-v4.4/product-guide/dragen-v4.4/dragen-dna-pipeline/sv-calling/sv-igv-tutorial#translocations
https://github.com/igvteam/igv/issues/1081

--- title: "Visualizing RNA-seq alignments with fusions with IGV" date: "2026-08-14" published-title: Created date-modified: last-modified categories: [RNA-seq, DESeq2, EdgeR, EBSeq] toc-title: "Contents" description: "It reconstructs the continuous paths that cells take as they change from one state to another." format: html: code-tools: source: true toggle: true execute: eval: false --- ## Step 1: Set Up Split-Screen View in IGV Fusions involve two distinct genomic loci (e.g., TMPRSS2 on Chr 21 and ERG on Chr 21, or two different chromosomes). 1. Go to the IGV search bar at the top. 2. Enter both gene locations separated by a space (e.g., TMPRSS2 ERG or chr21:41465000 chr21:38400000). 3. Press Enter. IGV will create a split panel view showing Gene 1 on the left and Gene 2 on the right. ## Step 2: Configure IGV Display Settings Right-click on the BAM track coverage/alignment panel and make sure the following settings are enabled: - Color alignments by → SA tag (already done) or pair orientation / insert size (crucial for pair-end discordant reads). - View as pairs: Enable this so paired-end reads are visually connected by a thin line. - Show soft-clipped bases: Enable this (Right-click track → Preferences → Alignments → check Show soft-clipped bases). ## Step 3: Identify & Count the Evidence ### A. Split Reads (Direct Junction Overlap) These are single sequencing reads that span across the fusion breakpoint. STAR aligns one part to Gene 1 and soft-clips the remaining sequence, which maps to Gene 2 as a supplementary alignment. - Visual Signature: - The read will display colored soft-clipped bases at the exact junction coordinate reported by Arriba.Right-click a split read and choose "View mate in split screen" or "Go to mate / supplementary alignment". The other portion of the read will highlight at the exact junction coordinate of Gene 2. - How to Count: Count every unique read ID that has a soft-clipped tail precisely terminating at the breakpoint position. - ### B. Discordant Mates (Spanning Pairs) These are paired-end reads where Read 1 maps entirely within Gene 1, and Read 2 maps entirely within Gene 2. - Visual Signature: - When set to Color alignments by insert size/pair orientation, these pairs will turn colored (usually teal, blue, or red instead of standard gray) because their insert size or chromosome mapping is abnormal. - Dragging across the split screen will show thin lines connecting Read 1 in Gene 1 to Read 2 in Gene 2 across the panel break. - How to Count: Count the number of uniquely colored pair connections bridging Gene 1 and Gene 2. ## Step 4: Validate Against Arriba/STAR-fusion Outputs Compare your visual count to the metrics in Arriba or STAR-Fusion’s output file (fusions.tsv): - Check Breakpoints: Verify that the soft-clip boundaries match the breakpoint1 and breakpoint2 genomic coordinates in fusions.tsv. - Compare Read Counts:split_reads1 / split_reads2: Should equal the number of split reads you count anchoring at Gene 1 and Gene 2 respectively. - discordant_mates: Should equal the count of uniquely mapped paired-end fragments bridging the two genes. ## Step 5: Rule Out Artifacts (False Positive Filters) To ensure the fusion is real and not an alignment artifact:Mapping Quality (MAPQ): - Hover over the supporting reads. Make sure the MAPQ is high (>20–30). - Soft-clipped reads with MAPQ 0 are usually caused by repetitive elements or homologous sequences. - Anchor Length: Ensure the split-read segments mapped to both sides of the junction have a reasonable anchor length (>15–20 bp), rather than just a few soft-clipped nucleotides. - Strand Orientation: Check if the reads align continuously with respect to the strand logic (e.g., sense-to-sense for exon-exon fusions). - Unmapped Mates: Beware if the region contains a high density of multi-mapping reads or unaligned sequences (colored purple/white mismatch bars), which points toward library preparation or mapping artifacts. ## References - https://okko73313.blogspot.com/2014/10/visualizing-rna-seq-alignments-with.html - https://help.connected.illumina.com/dragen/dragen-v4.4/product-guide/dragen-v4.4/dragen-dna-pipeline/sv-calling/sv-igv-tutorial#translocations - https://github.com/igvteam/igv/issues/1081