Skip to main content

Gene fusion calls and the DNA behind them

TL;DR: a fusion caller hands you a table of gene pairs and a junction coordinate, and nothing about the DNA event underneath. Load STAR-Fusion's table beside the long RNA reads it was called from, count the molecules that cross each junction, then find where the chromosome actually broke: for K562's BCR-ABL1 that is 122 kb away from the junction the caller reports, inside ABL1's first intron.

Prerequisites

The two python helpers are one file each:

curl -fO https://raw.githubusercontent.com/GMOD/jbrowse-components/main/scripts/depmap_to_jbrowse.py
curl -fO https://raw.githubusercontent.com/GMOD/jbrowse-components/main/scripts/lift_bnd_vcf.py

Where the data comes from

Four PacBio Iso-Seq runs from ENCODE, two tables out of DepMap's 24Q4 release, and the DNA breakpoints on hg19:

K562

K562 is a chronic myeloid leukemia line carrying the Philadelphia chromosome, the t(9;22) that fuses BCR to ABL1. Its transcripts here are long RNA reads, its fusion calls come out of DepMap's pipeline over short ones, and its DNA breakpoints come from a linked-read run, so the transcript answer and the DNA answer are three different assays apart.

Both DepMap tables cover every line in the release, so each starts with a row filter. depmap_to_jbrowse.py does that filter and writes what JBrowse reads: a STAR-Fusion TSV out of the fusion table, and a bedGraph out of the copy-number segments, which becomes the copy-number lane. K562 is model ACH-000551, and its WGS copy-number profile is PR-aheaZL:

python3 depmap_to_jbrowse.py fusions OmicsFusionFiltered.csv ACH-000551 K562.star-fusion.tsv
python3 depmap_to_jbrowse.py segments OmicsCNSegmentsProfile.csv PR-aheaZL K562_cn.bedGraph
sort -k1,1 -k2,2n K562_cn.bedGraph |
  awk 'NR==FNR{ok[$1];next} ($1 in ok)' hg38.chrom.sizes - > K562_cn.sorted.bedGraph
bedGraphToBigWig K562_cn.sorted.bedGraph hg38.chrom.sizes K562_cn.bw

The DNA breakpoints arrive on hg19, and lifting a BND callset is the one step here that fails quietly. A breakend record carries a second coordinate inside its ALT string, so a plain liftOver of the POS column produces a valid VCF whose partner coordinates still point at hg19. lift_bnd_vcf.py moves both:

python3 lift_bnd_vcf.py calls.hg19.vcf.gz hg19ToHg38.over.chain.gz \
  ./liftOver calls.hg38.vcf liftwork
bgzip calls.hg38.vcf && tabix -p vcf calls.hg38.vcf.gz

The script's five arguments are the input VCF, the chain, the liftOver binary, the output VCF and a scratch directory.

Triaging the calls

The SV inspector opens DepMap's STAR-Fusion output as a table beside a circular view of it, one chord per row. Add → SV inspector, then a File Type of STAR-Fusion, which the wizard cannot infer from this file's .tsv extension. Both of those steps are live links under the figure below.

Searching the table narrows both halves, since the circle draws the rows the search leaves. chr9 leaves BCR--ABL1 and NUP214--XKR3, one junction seen from both sides, carrying more junction reads than anything else in the file.

Every row carries a menu on its caret, and Open in linear genome view takes it to its own breakpoint; type the partner's window into the location box after it and the view holds both side by side.

Then turn on Read connections → View as pairs. That merges each molecule's two alignments onto one row across the two regions, so the fusion reads as a flat line per molecule. Flip the chr22 region as well ([rev]), since XKR3 is on the minus strand and the flip runs both halves of a molecule the same way.

NUP214--XKR3 as two regions of one view with reads linked, opened from its row in the SV inspector. The breakpoints are banded green and each line is one Iso-Seq molecule running from NUP214 into XKR3. Open in JBrowse: Import form ↗ · All 44 calls ↗ · Searched for chr9 ↗ · Linked reads ↗ (large dataset — this session takes a while to load)

The steps behind the figure above, run against your own data.

  1. Open your genome: on the JBrowse Desktop start screen click Open new genome (or Show all available genomes to pick a hosted one).This figure uses hg38.
  2. Click the view's title in its header and enter "Add → SV inspector, File Type → STAR-Fusion"
  3. Import form → File Type → STAR-FusionOnly needed when the filename does not end in the type, which is why a STAR-Fusion `.tsv` has to be named and a `.vcf.gz` does not.
  4. Drag the bar at the bottom edge of the view to resize it (620px here).

Open this view in JBrowse Desktop ↗

Opens JBrowse Desktop (JBrowse Desktop 5.0+) at this view and saves it as a reopenable session — swap in your own files afterwards.

Nothing happens? Paste this link into Desktop's Open JBrowse Web link... (start screen, or File → Session):

https://jbrowse.org/code/jb2/main/?config=https%3A%2F%2Fjbrowse.org%2Fdemos%2Fcancer_sv%2Fconfig.json&session=spec-%7B%22views%22%3A%5B%7B%22type%22%3A%22SvInspectorView%22%2C%22displayName%22%3A%22Add+%E2%86%92+SV+inspector%2C+File+Type+%E2%86%92+STAR-Fusion%22%2C%22assembly%22%3A%22hg38%22%2C%22fileType%22%3A%22STAR-Fusion%22%2C%22height%22%3A620%7D%5D%7D&sessionName=JBrowse+docs%3A+cancer_sv%2Fk562_fusion_inspector_form

This session spec pastes after &session=spec- on any JBrowse Web instance.

{
  "views": [
    {
      "type": "SvInspectorView",
      "displayName": "Add → SV inspector, File Type → STAR-Fusion",
      "assembly": "hg38",
      "fileType": "STAR-Fusion",
      "height": 620
    }
  ]
}

Loaded against config: https://jbrowse.org/demos/cancer_sv/config.json

Hand this to a coding agent. It rebuilds the figure above headlessly, and the session file is then the thing to edit — swap an adapter uri for your own file and rerun.

cat > session.json <<'JSON'
{
  "views": [
    {
      "type": "SvInspectorView",
      "displayName": "Add → SV inspector, File Type → STAR-Fusion",
      "assembly": "hg38",
      "fileType": "STAR-Fusion",
      "height": 620
    }
  ]
}
JSON

npx @jbrowse/capture --instance https://jbrowse.org/code/jb2/main/ \
  --config https://jbrowse.org/demos/cancer_sv/config.json \
  --session session.json -o figure.png

Using JBrowse with AI agents covers the rest: hosted genomes to point it at, and why waiting for the render is the part that goes wrong.

The steps behind the figure above, run against your own data.

  1. Open your genome: on the JBrowse Desktop start screen click Open new genome (or Show all available genomes to pick a hosted one).This figure uses hg38.
  2. Click the view's title in its header and enter "All 44 STAR-Fusion calls"
  3. Import form → Open file from URL or local computer → URL → Enter URL → OpenOpen the form with Add → SV inspector (or Add → Spreadsheet view); set File Type and Assembly in the same form before clicking Open.
  4. Import form → File Type → STAR-FusionOnly needed when the filename does not end in the type, which is why a STAR-Fusion `.tsv` has to be named and a `.vcf.gz` does not.
  5. Drag the bar at the bottom edge of the view to resize it (620px here).

Open this view in JBrowse Desktop ↗

Opens JBrowse Desktop (JBrowse Desktop 5.0+) at this view and saves it as a reopenable session — swap in your own files afterwards.

Nothing happens? Paste this link into Desktop's Open JBrowse Web link... (start screen, or File → Session):

https://jbrowse.org/code/jb2/main/?config=https%3A%2F%2Fjbrowse.org%2Fdemos%2Fcancer_sv%2Fconfig.json&session=spec-%7B%22views%22%3A%5B%7B%22type%22%3A%22SvInspectorView%22%2C%22displayName%22%3A%22All+44+STAR-Fusion+calls%22%2C%22assembly%22%3A%22hg38%22%2C%22uri%22%3A%22https%3A%2F%2Fjbrowse.org%2Fdemos%2Fcancer_sv%2FK562.star-fusion.tsv%22%2C%22fileType%22%3A%22STAR-Fusion%22%2C%22height%22%3A620%7D%5D%7D&sessionName=JBrowse+docs%3A+cancer_sv%2Fk562_fusion_inspector_all

This session spec pastes after &session=spec- on any JBrowse Web instance.

{
  "views": [
    {
      "type": "SvInspectorView",
      "displayName": "All 44 STAR-Fusion calls",
      "assembly": "hg38",
      "uri": "https://jbrowse.org/demos/cancer_sv/K562.star-fusion.tsv",
      "fileType": "STAR-Fusion",
      "height": 620
    }
  ]
}

Loaded against config: https://jbrowse.org/demos/cancer_sv/config.json

Hand this to a coding agent. It rebuilds the figure above headlessly, and the session file is then the thing to edit — swap an adapter uri for your own file and rerun.

cat > session.json <<'JSON'
{
  "views": [
    {
      "type": "SvInspectorView",
      "displayName": "All 44 STAR-Fusion calls",
      "assembly": "hg38",
      "uri": "https://jbrowse.org/demos/cancer_sv/K562.star-fusion.tsv",
      "fileType": "STAR-Fusion",
      "height": 620
    }
  ]
}
JSON

npx @jbrowse/capture --instance https://jbrowse.org/code/jb2/main/ \
  --config https://jbrowse.org/demos/cancer_sv/config.json \
  --session session.json -o figure.png

Using JBrowse with AI agents covers the rest: hosted genomes to point it at, and why waiting for the render is the part that goes wrong.

The steps behind the figure above, run against your own data.

  1. Open your genome: on the JBrowse Desktop start screen click Open new genome (or Show all available genomes to pick a hosted one).This figure uses hg38.
  2. Click the view's title in its header and enter "Search "chr9": the reciprocal pair"
  3. Import form → Open file from URL or local computer → URL → Enter URL → OpenOpen the form with Add → SV inspector (or Add → Spreadsheet view); set File Type and Assembly in the same form before clicking Open.
  4. Import form → File Type → STAR-FusionOnly needed when the filename does not end in the type, which is why a STAR-Fusion `.tsv` has to be named and a `.vcf.gz` does not.
  5. Type "chr9" into the table's search box (the magnifier above the grid)In the SV inspector the circular view draws only the rows the search leaves, so this narrows the chords too — it does not change which regions the circle shows, which is the 'show only regions with data' checkbox beside it.
  6. Drag the bar at the bottom edge of the view to resize it (620px here).

Open this view in JBrowse Desktop ↗

Opens JBrowse Desktop (JBrowse Desktop 5.0+) at this view and saves it as a reopenable session — swap in your own files afterwards.

Nothing happens? Paste this link into Desktop's Open JBrowse Web link... (start screen, or File → Session):

https://jbrowse.org/code/jb2/main/?config=https%3A%2F%2Fjbrowse.org%2Fdemos%2Fcancer_sv%2Fconfig.json&session=spec-%7B%22views%22%3A%5B%7B%22type%22%3A%22SvInspectorView%22%2C%22displayName%22%3A%22Search+%5C%22chr9%5C%22%3A+the+reciprocal+pair%22%2C%22assembly%22%3A%22hg38%22%2C%22uri%22%3A%22https%3A%2F%2Fjbrowse.org%2Fdemos%2Fcancer_sv%2FK562.star-fusion.tsv%22%2C%22fileType%22%3A%22STAR-Fusion%22%2C%22filterText%22%3A%22chr9%22%2C%22height%22%3A620%7D%5D%7D&sessionName=JBrowse+docs%3A+cancer_sv%2Fk562_fusion_inspector_pair

This session spec pastes after &session=spec- on any JBrowse Web instance.

{
  "views": [
    {
      "type": "SvInspectorView",
      "displayName": "Search \"chr9\": the reciprocal pair",
      "assembly": "hg38",
      "uri": "https://jbrowse.org/demos/cancer_sv/K562.star-fusion.tsv",
      "fileType": "STAR-Fusion",
      "filterText": "chr9",
      "height": 620
    }
  ]
}

Loaded against config: https://jbrowse.org/demos/cancer_sv/config.json

Hand this to a coding agent. It rebuilds the figure above headlessly, and the session file is then the thing to edit — swap an adapter uri for your own file and rerun.

cat > session.json <<'JSON'
{
  "views": [
    {
      "type": "SvInspectorView",
      "displayName": "Search \"chr9\": the reciprocal pair",
      "assembly": "hg38",
      "uri": "https://jbrowse.org/demos/cancer_sv/K562.star-fusion.tsv",
      "fileType": "STAR-Fusion",
      "filterText": "chr9",
      "height": 620
    }
  ]
}
JSON

npx @jbrowse/capture --instance https://jbrowse.org/code/jb2/main/ \
  --config https://jbrowse.org/demos/cancer_sv/config.json \
  --session session.json -o figure.png

Using JBrowse with AI agents covers the rest: hosted genomes to point it at, and why waiting for the render is the part that goes wrong.

The steps behind the figure above, run against your own data.

  1. Open your genome: on the JBrowse Desktop start screen click Open new genome (or Show all available genomes to pick a hosted one).This figure uses hg38.
  2. Add your own track: File → Open track..., then paste a URL or choose a local file.This figure uses the “ncbi_refseq_hg38” track.
  3. Track menu → Gene glyph → Representative transcript“ncbi_refseq_hg38”.
  4. Drag the bar at the bottom edge of the track to resize it.“ncbi_refseq_hg38”.
  5. Add your own track: File → Open track..., then paste a URL or choose a local file.This figure uses the “K562_star_fusion” track.
  6. Track menu → Set feature height → Collapsed“K562_star_fusion”.
  7. Drag the bar at the bottom edge of the track to resize it.“K562_star_fusion”.
  8. Add your own track: File → Open track..., then paste a URL or choose a local file.This figure uses the “K562_isoseq” track.
  9. Drag the bar at the bottom edge of the track to resize it.“K562_isoseq”.
  10. Drag the bottom edge of the coverage band to resize it.“K562_isoseq”.
  11. Track menu → Filter by... → Split alignments → Only split alignmentsA read filter drops reads in the worker before layout, so it thins the coverage histogram along with the pileup.“K562_isoseq”.
  12. Track menu → Read connections → View as pairs / link supplementary alignments (checked)“K562_isoseq”.
  13. Track menu → Show... → Show legend (checked)“K562_isoseq”.
  14. Track menu → Read connections → Use curved connectors (checked)Draws a read's alignments joined by a curve rather than a straight line. Worth it when the two ends sit in different displayed regions, where the curve reads as one read crossing the gap.“K562_isoseq”.
  15. Track menu → Read connections → Show read arcs“K562_isoseq”.
  16. Drag the handle at the edge of the arc band to resize it (90px here).Its tooltip reads "Drag to resize arcs area"; there is no menu entry for the height.“K562_isoseq”.
  17. Track menu → Read height → Custom... → 4px“K562_isoseq”.
  18. Type your region of interest into the location box and press Enter.This figure is at chr9:131,186,500-131,199,515 chr22:16,795,568-16,808,583[rev].
  19. View menu → Zoom to region / use the location box, then add a highlightThis figure highlights 2 regions; highlights can also be set with the &highlight= URL parameter.

Open this view in JBrowse Desktop ↗

Opens JBrowse Desktop (JBrowse Desktop 5.0+) at this view and saves it as a reopenable session — swap in your own files afterwards.

Nothing happens? Paste this link into Desktop's Open JBrowse Web link... (start screen, or File → Session):

https://jbrowse.org/code/jb2/main/?config=https%3A%2F%2Fjbrowse.org%2Fdemos%2Fcancer_sv%2Fconfig.json&session=spec-%7B%22views%22%3A%5B%7B%22type%22%3A%22LinearGenomeView%22%2C%22assembly%22%3A%22hg38%22%2C%22loc%22%3A%22chr9%3A131%2C186%2C500-131%2C199%2C515+chr22%3A16%2C795%2C568-16%2C808%2C583%5Brev%5D%22%2C%22highlight%22%3A%5B%7B%22refName%22%3A%22chr9%22%2C%22start%22%3A131198915%2C%22end%22%3A131199115%7D%2C%7B%22refName%22%3A%22chr22%22%2C%22start%22%3A16807983%2C%22end%22%3A16808183%7D%5D%2C%22tracks%22%3A%5B%7B%22trackId%22%3A%22ncbi_refseq_hg38%22%2C%22geneGlyphMode%22%3A%22longestCoding%22%2C%22height%22%3A60%7D%2C%7B%22trackId%22%3A%22K562_star_fusion%22%2C%22type%22%3A%22LinearVariantDisplay%22%2C%22displayMode%22%3A%22collapsed%22%2C%22height%22%3A30%7D%2C%7B%22trackId%22%3A%22K562_isoseq%22%2C%22type%22%3A%22LinearAlignmentsDisplay%22%2C%22height%22%3A600%2C%22coverageHeight%22%3A150%2C%22filterBy%22%3A%7B%22split%22%3A%22only%22%7D%2C%22linkedReads%22%3A%22normal%22%2C%22showLegend%22%3Atrue%2C%22showBezierConnections%22%3Atrue%2C%22readConnections%22%3A%22arc%22%2C%22readConnectionsHeight%22%3A90%2C%22featureHeight%22%3A4%7D%5D%7D%5D%7D&sessionName=JBrowse+docs%3A+cancer_sv%2Fk562_fusion_inspector_reads

This session spec pastes after &session=spec- on any JBrowse Web instance.

{
  "views": [
    {
      "type": "LinearGenomeView",
      "assembly": "hg38",
      "loc": "chr9:131,186,500-131,199,515 chr22:16,795,568-16,808,583[rev]",
      "highlight": [
        {
          "refName": "chr9",
          "start": 131198915,
          "end": 131199115
        },
        {
          "refName": "chr22",
          "start": 16807983,
          "end": 16808183
        }
      ],
      "tracks": [
        {
          "trackId": "ncbi_refseq_hg38",
          "geneGlyphMode": "longestCoding",
          "height": 60
        },
        {
          "trackId": "K562_star_fusion",
          "type": "LinearVariantDisplay",
          "displayMode": "collapsed",
          "height": 30
        },
        {
          "trackId": "K562_isoseq",
          "type": "LinearAlignmentsDisplay",
          "height": 600,
          "coverageHeight": 150,
          "filterBy": {
            "split": "only"
          },
          "linkedReads": "normal",
          "showLegend": true,
          "showBezierConnections": true,
          "readConnections": "arc",
          "readConnectionsHeight": 90,
          "featureHeight": 4
        }
      ]
    }
  ]
}

Loaded against config: https://jbrowse.org/demos/cancer_sv/config.json

The same view with jbrowse-anywidget. Point the adapter at your own file.

from jbrowse_anywidget import LinearGenomeView, fetch_hub

view = LinearGenomeView(
    assembly=fetch_hub("hg38"),
    location="chr9:131,186,500-131,199,515 chr22:16,795,568-16,808,583[rev]",
)

view.add_track({
    "type": "FeatureTrack",
    "trackId": "my_track_1",
    "name": "My track 1",
    "assemblyNames": ["hg38"],
    "adapter": {"type": "BamAdapter", "uri": "https://your-server/your-file"},
})
view.add_track({
    "type": "FeatureTrack",
    "trackId": "my_track_2",
    "name": "My track 2",
    "assemblyNames": ["hg38"],
    "adapter": {"type": "BamAdapter", "uri": "https://your-server/your-file"},
})
view.add_track({
    "type": "FeatureTrack",
    "trackId": "my_track_3",
    "name": "My track 3",
    "assemblyNames": ["hg38"],
    "adapter": {"type": "BamAdapter", "uri": "https://your-server/your-file"},
})

view  # display the widget

Hand this to a coding agent. It rebuilds the figure above headlessly, and the session file is then the thing to edit — swap an adapter uri for your own file and rerun.

cat > session.json <<'JSON'
{
  "views": [
    {
      "type": "LinearGenomeView",
      "assembly": "hg38",
      "loc": "chr9:131,186,500-131,199,515 chr22:16,795,568-16,808,583[rev]",
      "highlight": [
        {
          "refName": "chr9",
          "start": 131198915,
          "end": 131199115
        },
        {
          "refName": "chr22",
          "start": 16807983,
          "end": 16808183
        }
      ],
      "tracks": [
        {
          "trackId": "ncbi_refseq_hg38",
          "geneGlyphMode": "longestCoding",
          "height": 60
        },
        {
          "trackId": "K562_star_fusion",
          "type": "LinearVariantDisplay",
          "displayMode": "collapsed",
          "height": 30
        },
        {
          "trackId": "K562_isoseq",
          "type": "LinearAlignmentsDisplay",
          "height": 600,
          "coverageHeight": 150,
          "filterBy": {
            "split": "only"
          },
          "linkedReads": "normal",
          "showLegend": true,
          "showBezierConnections": true,
          "readConnections": "arc",
          "readConnectionsHeight": 90,
          "featureHeight": 4
        }
      ]
    }
  ]
}
JSON

npx @jbrowse/capture --instance https://jbrowse.org/code/jb2/main/ \
  --config https://jbrowse.org/demos/cancer_sv/config.json \
  --session session.json -o figure.png

Using JBrowse with AI agents covers the rest: hosted genomes to point it at, and why waiting for the render is the part that goes wrong.

That is the NUP214--XKR3 side of the pair, and BCR--ABL1 takes the rest of this page in the same layout, off the STAR-Fusion calls the build script adds as this track:

{
  "type": "VariantTrack",
  "trackId": "K562_star_fusion",
  "name": "K562 STAR-Fusion calls (DepMap 24Q4)",
  "assemblyNames": ["hg38"],
  "adapter": {
    "type": "StarFusionAdapter",
    "starFusionLocation": { "uri": "K562.star-fusion.tsv" }
  }
}

BCR-ABL1 across three regions

The Iso-Seq reads stop and start at the bases STAR-Fusion reported from short reads. Putting both partners in one view as displayed regions lays the fusion out the way FusionInspector does: right-click a read that crosses the junction and choose Split current view to show split alignments, or type the locations into the location box, separated by spaces. The transcript reaches ABL1 at more than one place, so this view uses three regions, the BCR donor and two acceptor windows.

A read that crosses the junction is one alignment on chr22 and a supplementary alignment on chr9. Read connections → Use curved connectors draws a curve between the two, and with both partners displayed those curves cross from one region into the other. Filter by... → Split alignments → Only split alignments then drops every read that stays on one chromosome, so the pileup is the fusion's own support.

Near-identical curves stack into one line, so a curve per molecule cannot show how many molecules agree on a junction. Read connections → Show read arcs adds a band under the coverage where each junction is drawn once, thickened by the reads behind it. An arc is drawn when both of its ends are in view, so it reaches across a region divider, and each acceptor window receives one. The vertical at the BCR donor stands for the molecules whose ABL1 alignment lands in neither window.

BCR on chr22 beside two ABL1 windows on chr9 as three regions of one view, showing only split reads with supplementary alignments linked. The arc band draws one counted arc from the BCR donor into each ABL1 window, and only the right-hand window carries a STAR-Fusion band. Open this view in JBrowse ↗

The steps behind the figure above, run against your own data.

  1. Open your genome: on the JBrowse Desktop start screen click Open new genome (or Show all available genomes to pick a hosted one).This figure uses hg38.
  2. Add your own track: File → Open track..., then paste a URL or choose a local file.This figure uses the “ncbi_refseq_hg38” track.
  3. Track menu → Gene glyph → Representative transcript“ncbi_refseq_hg38”.
  4. Drag the bar at the bottom edge of the track to resize it.“ncbi_refseq_hg38”.
  5. Add your own track: File → Open track..., then paste a URL or choose a local file.This figure uses the “K562_isoseq” track.
  6. Track menu → Read height → Track sizing → Fixed read height + autogrow track height“K562_isoseq”.
  7. Drag the bottom edge of the coverage band to resize it.“K562_isoseq”.
  8. Track menu → Filter by... → Split alignments → Only split alignmentsA read filter drops reads in the worker before layout, so it thins the coverage histogram along with the pileup.“K562_isoseq”.
  9. Track menu → Read connections → View as pairs / link supplementary alignments (checked)“K562_isoseq”.
  10. Track menu → Read connections → Use curved connectors (checked)Draws a read's alignments joined by a curve rather than a straight line. Worth it when the two ends sit in different displayed regions, where the curve reads as one read crossing the gap.“K562_isoseq”.
  11. Track menu → Read connections → Show read arcs“K562_isoseq”.
  12. Drag the handle at the edge of the arc band to resize it (90px here).Its tooltip reads "Drag to resize arcs area"; there is no menu entry for the height.“K562_isoseq”.
  13. Track menu → Show... → Show legend (checked)“K562_isoseq”.
  14. Track menu → Read height → Custom... → 4px“K562_isoseq”.
  15. Type your region of interest into the location box and press Enter.This figure is at chr22:23,286,000-23,293,000 chr9:130,778,000-130,785,000 chr9:130,851,000-130,858,000.
  16. View menu → Zoom to region / use the location box, then add a highlightThis figure highlights 2 regions; highlights can also be set with the &highlight= URL parameter.

Some settings have no written step yet — see the Session spec tab.

Open this view in JBrowse Desktop ↗

Opens JBrowse Desktop (JBrowse Desktop 5.0+) at this view and saves it as a reopenable session — swap in your own files afterwards.

Nothing happens? Paste this link into Desktop's Open JBrowse Web link... (start screen, or File → Session):

https://jbrowse.org/code/jb2/main/?config=https%3A%2F%2Fjbrowse.org%2Fdemos%2Fcancer_sv%2Fconfig.json&session=spec-%7B%22views%22%3A%5B%7B%22type%22%3A%22LinearGenomeView%22%2C%22assembly%22%3A%22hg38%22%2C%22loc%22%3A%22chr22%3A23%2C286%2C000-23%2C293%2C000+chr9%3A130%2C778%2C000-130%2C785%2C000+chr9%3A130%2C851%2C000-130%2C858%2C000%22%2C%22highlight%22%3A%5B%7B%22refName%22%3A%22chr22%22%2C%22start%22%3A23290313%2C%22end%22%3A23290513%7D%2C%7B%22refName%22%3A%22chr9%22%2C%22start%22%3A130853964%2C%22end%22%3A130854164%7D%5D%2C%22tracks%22%3A%5B%7B%22trackId%22%3A%22ncbi_refseq_hg38%22%2C%22geneGlyphMode%22%3A%22longestCoding%22%2C%22height%22%3A90%7D%2C%7B%22trackId%22%3A%22K562_isoseq%22%2C%22type%22%3A%22LinearAlignmentsDisplay%22%2C%22heightMode%22%3A%22grow%22%2C%22growMaxHeight%22%3A2400%2C%22coverageHeight%22%3A190%2C%22filterBy%22%3A%7B%22split%22%3A%22only%22%7D%2C%22linkedReads%22%3A%22normal%22%2C%22showBezierConnections%22%3Atrue%2C%22readConnections%22%3A%22arc%22%2C%22readConnectionsHeight%22%3A90%2C%22showLegend%22%3Atrue%2C%22featureHeight%22%3A4%7D%5D%7D%5D%7D&sessionName=JBrowse+docs%3A+cancer_sv%2Fk562_bcr_abl_split

This session spec pastes after &session=spec- on any JBrowse Web instance.

{
  "views": [
    {
      "type": "LinearGenomeView",
      "assembly": "hg38",
      "loc": "chr22:23,286,000-23,293,000 chr9:130,778,000-130,785,000 chr9:130,851,000-130,858,000",
      "highlight": [
        {
          "refName": "chr22",
          "start": 23290313,
          "end": 23290513
        },
        {
          "refName": "chr9",
          "start": 130853964,
          "end": 130854164
        }
      ],
      "tracks": [
        {
          "trackId": "ncbi_refseq_hg38",
          "geneGlyphMode": "longestCoding",
          "height": 90
        },
        {
          "trackId": "K562_isoseq",
          "type": "LinearAlignmentsDisplay",
          "heightMode": "grow",
          "growMaxHeight": 2400,
          "coverageHeight": 190,
          "filterBy": {
            "split": "only"
          },
          "linkedReads": "normal",
          "showBezierConnections": true,
          "readConnections": "arc",
          "readConnectionsHeight": 90,
          "showLegend": true,
          "featureHeight": 4
        }
      ]
    }
  ]
}

Loaded against config: https://jbrowse.org/demos/cancer_sv/config.json

The same view with jbrowse-anywidget. Point the adapter at your own file.

from jbrowse_anywidget import LinearGenomeView, fetch_hub

view = LinearGenomeView(
    assembly=fetch_hub("hg38"),
    location="chr22:23,286,000-23,293,000 chr9:130,778,000-130,785,000 chr9:130,851,000-130,858,000",
)

view.add_track({
    "type": "FeatureTrack",
    "trackId": "my_track_1",
    "name": "My track 1",
    "assemblyNames": ["hg38"],
    "adapter": {"type": "BamAdapter", "uri": "https://your-server/your-file"},
})
view.add_track({
    "type": "FeatureTrack",
    "trackId": "my_track_2",
    "name": "My track 2",
    "assemblyNames": ["hg38"],
    "adapter": {"type": "BamAdapter", "uri": "https://your-server/your-file"},
})

view  # display the widget

Hand this to a coding agent. It rebuilds the figure above headlessly, and the session file is then the thing to edit — swap an adapter uri for your own file and rerun.

cat > session.json <<'JSON'
{
  "views": [
    {
      "type": "LinearGenomeView",
      "assembly": "hg38",
      "loc": "chr22:23,286,000-23,293,000 chr9:130,778,000-130,785,000 chr9:130,851,000-130,858,000",
      "highlight": [
        {
          "refName": "chr22",
          "start": 23290313,
          "end": 23290513
        },
        {
          "refName": "chr9",
          "start": 130853964,
          "end": 130854164
        }
      ],
      "tracks": [
        {
          "trackId": "ncbi_refseq_hg38",
          "geneGlyphMode": "longestCoding",
          "height": 90
        },
        {
          "trackId": "K562_isoseq",
          "type": "LinearAlignmentsDisplay",
          "heightMode": "grow",
          "growMaxHeight": 2400,
          "coverageHeight": 190,
          "filterBy": {
            "split": "only"
          },
          "linkedReads": "normal",
          "showBezierConnections": true,
          "readConnections": "arc",
          "readConnectionsHeight": 90,
          "showLegend": true,
          "featureHeight": 4
        }
      ]
    }
  ]
}
JSON

npx @jbrowse/capture --instance https://jbrowse.org/code/jb2/main/ \
  --config https://jbrowse.org/demos/cancer_sv/config.json \
  --session session.json -o figure.png

Using JBrowse with AI agents covers the rest: hosted genomes to point it at, and why waiting for the render is the part that goes wrong.

Where the DNA broke

BCR-ABL1 is amplified as well as expressed. Both chr9 breakpoints fall inside a segment at roughly seven copies, while the chr22 partners sit at one, so what is amplified is the piece of chr9 that the two junctions cut out. DepMap's segmentation covers no interval over BCR itself, which is why that window has an arc but no copy-number step under it.

A fusion caller only reports junctions that are transcribed, so those arcs land on exon boundaries and cannot say where the amplified block begins. The DNA answer comes from a different assay: ENCODE's 10X Chromium linked-read run on K562 (ENCSR053AXS), whose large-SV calls are on hg19 and are lifted to hg38 by the build script. Its chr9 breakpoint for BCR-ABL1 is at 130,731,760, and DepMap's copy-number segmentation steps up at 130,731,326. The transcript junction is 122 kb to the right of both, inside ABL1's first intron: the amplicon boundary is a DNA break, and the transcript is spliced from it to the nearest exon.

This is the reasoning SplitThreader applied to the ERBB2 amplicon in SK-BR-3 (Nattestad et al. 2018): copy-number steps and breakpoints that describe the same interval are evidence of one event. Here two independent assays put that interval's edge in the same place.

Reproduce it end to end

scripts/build_cancer_sv_demo.sh builds everything above from public sources:

curl -fO https://raw.githubusercontent.com/GMOD/jbrowse-components/main/scripts/build_cancer_sv_demo.sh
bash build_cancer_sv_demo.sh    # builds ./cancer_sv_build/jbrowse2
npx --yes serve cancer_sv_build/jbrowse2

It pulls the four ENCODE Iso-Seq alignments and merges them, converts the DepMap release into a STAR-Fusion TSV and a copy-number bigWig, and lifts the ENCODE linked-read breakpoints onto hg38. The same run builds the COLO829 half of the demo, which Complex rearrangements and derivative alleles walks through.

See also

References

  • Nattestad M, et al. Complex rearrangements and oncogene amplifications revealed by long-read DNA and RNA sequencing of a breast cancer cell line. Genome Research (2018). https://doi.org/10.1101/gr.231100.117