SPI1 Binding-QTL Dynseq Track¶

This example compares base-resolution sequence contribution scores for the reference C and alternate G alleles of the SPI1 binding QTL rs5764238 in GM12878. Letter height represents the projected DeepSHAP contribution score from the SPI1 ChIP-seq BPNet model; negative scores extend below zero. It recreates the reference-versus-alternate SPI1 ChIP plot in the dynseq SPI1 bQTL notebook.

{
  "description": [
    "SPI1 Binding-QTL Dynseq Track",
    "Compares per-base projected DeepSHAP contribution scores for the reference C and alternate G alleles of rs5764238 in GM12878 SPI1 ChIP-seq BPNet predictions.",
    "Data source: pinned original BigWigs from the dynseq-paper SPI1 bQTL vignette. Terms: Zenodo record 6582100 is licensed under CC BY 4.0."
  ],

  "assembly": "hg38",

  "scales": {
    "x": {
      "domain": [
        { "chrom": "chr22", "pos": 43720850 },
        { "chrom": "chr22", "pos": 43720960 }
      ],
      "zoom": {
        "extent": [
          { "chrom": "chr22", "pos": 43719872 },
          { "chrom": "chr22", "pos": 43721985 }
        ]
      }
    }
  },

  "data": {
    "lazy": {
      "type": "indexedFasta",
      "url": "https://data.genomespy.app/genomes/hg38/hg38.fa"
    }
  },

  "transform": [
    {
      "type": "flattenSequence",
      "field": "sequence",
      "as": ["rawPos", "base"]
    },
    { "type": "formula", "expr": "upper(datum.base)", "as": "base" },
    { "type": "formula", "expr": "datum.start + datum.rawPos", "as": "pos" }
  ],

  "resolve": { "scale": { "y": "shared" }, "axis": { "x": "shared" } },

  "vconcat": [
    {
      "import": { "template": "allele-track" },
      "params": { "allele": "ref" }
    },
    {
      "import": { "template": "allele-track" },
      "params": { "allele": "alt" }
    }
  ],

  "templates": {
    "allele-track": {
      "params": [{ "name": "allele", "value": "ref" }],

      "title": {
        "text": {
          "expr": "allele === 'ref' ? 'Reference allele (C)' : 'Alternate allele (G)'"
        },
        "style": "overlay-title"
      },
      "height": 120,

      "transform": [
        {
          "type": "coordinateLookup",
          "from": {
            "data": {
              "lazy": {
                "type": "bigwig",
                "url": {
                  "expr": "allele === 'ref' ? 'https://raw.githubusercontent.com/kundajelab/dynseq-paper/febc9180d72e92302d35c549002e0d56c79c536e/SPI1_bQTL/bigwigs/chip_imp_ref.bw' : 'https://raw.githubusercontent.com/kundajelab/dynseq-paper/febc9180d72e92302d35c549002e0d56c79c536e/SPI1_bQTL/bigwigs/chip_imp_alt.bw'"
                },
                "pixelsPerBin": 1
              }
            },
            "transform": [
              { "type": "formula", "expr": "datum.start", "as": "pos" }
            ]
          },
          "key": ["chrom", "pos"],
          "values": ["score"]
        },
        { "type": "filter", "expr": "isValid(datum.score)" },
        {
          "type": "formula",
          "description": "Show the rs5764238 alternate G allele while retaining the shared reference FASTA source.",
          "expr": "allele === 'alt' && datum.pos === 43720929 ? 'G' : datum.base",
          "as": "base"
        }
      ],

      "layer": [
        {
          "name": "baseline",
          "data": { "values": [{}] },
          "encoding": {
            "y": { "datum": 0, "type": "quantitative" },
            "color": { "value": "gray" }
          },
          "mark": "rule"
        },
        {
          "name": "dynseq",
          "encoding": {
            "x": {
              "chrom": "chrom",
              "pos": "pos",
              "type": "locus"
            },
            "y": {
              "datum": 0,
              "type": "quantitative",
              "scale": { "zero": true, "nice": false },
              "axis": { "title": "Score" }
            },
            "y2": { "field": "score" },
            "text": { "field": "base" },
            "color": {
              "field": "base",
              "type": "nominal",
              "scale": {
                "domain": ["A", "C", "G", "T"],
                "range": ["green", "blue", "orange", "red"]
              },
              "legend": null
            },
            "tooltip": [
              { "field": "base", "type": "nominal" },
              { "field": "score", "type": "quantitative" }
            ]
          },
          "mark": {
            "type": "text",
            "font": "Source Sans Pro",
            "fontWeight": 700,
            "size": 100,
            "squeeze": true,
            "fitToBand": true,
            "paddingX": 0,
            "paddingY": 0,
            "logoLetters": true
          }
        }
      ]
    }
  },

  "config": { "view": { "fill": "#FAFAFA" } }
}

The visualization loads the original chip_imp_ref.bw and chip_imp_alt.bw directly from the pinned dynseq-paper source revision. The score values and missing-value gaps are not modified. The Zenodo source data are distributed under CC BY 4.0.

Source attribution: dynseq tracks data, Zenodo record 6582100; Nair et al., The dynseq browser track shows context-specific features at nucleotide resolution, Nature Genetics 54, 1581–1583 (2022); and Tehranchi et al., Pooled ChIP-Seq Links Variation in Transcription Factor Binding to Complex Disease Risk, Cell 165, 730–741 (2016).

What to notice¶

The reference bases come from a lazy indexed hg38 FASTA source and are expanded to uppercase, base-level rows once for both tracks. Each allele-track template then loads its corresponding BigWig and uses coordinateLookup to join scores to those bases. The alternate track displays G at rs5764238. The initial 110 bp view focuses on the variant and motif. You can pan or zoom out across the full 2,114 bp score window.

GenomeSpy Features¶

This example combines:

Lazy data sources for indexed FASTA and BigWig.
coordinateLookup to match base positions to BigWig score positions.
flattenSequence to expand reference-sequence chunks to base-level rows.
text marks with logoLetters to scale base characters between zero and their signed contribution scores.