genome-spy-python 0.6.0__tar.gz → 0.7.0__tar.gz

genome_spy_python-0.7.0/.agents/skills/commit/SKILL.md

@@ -0,0 +1,48 @@
+---
+name: commit
+description: Commit repository changes in small thematic chunks using conventional commits and concise changelog staging.
+---
+
+# Commit workflow
+
+Prefer small, independently understandable commits. Keep unrelated user changes
+unstaged. Do not amend, rewrite history, revert, force-push, or add co-author
+trailers unless explicitly requested.
+
+## Inspect and commit
+
+1. Start with `git status --short` and `git diff --stat`.
+2. Group changes into thematic slices. Split separate features, fixes, refactors,
+   tooling/configuration, documentation, and unrelated test work when separable;
+   do not split mechanically by file.
+3. Keep implementation with its direct tests, API/schema changes with required
+   supporting code, and generated files with their source change.
+4. Read targeted diffs only when needed, including when grouping is unclear:
+   `git diff -- <files>`. Minimize token use; do not read the full diff or re-review
+   code by default.
+5. Stage one slice, check `git diff --cached --stat`, and commit with a conventional
+   message. Repeat for remaining slices.
+6. Finish with `git status --short`, update changelog staging once if needed, and
+   report commit hashes and remaining uncommitted files.
+
+## Commit messages
+
+Use `<type>(<scope>): <description>`. Common types: `feat`, `fix`, `refactor`,
+`perf`, `test`, `docs`, `chore`, `build`, and `ci`.
+
+```text
+feat(parser): add strand-aware bigwig loading
+fix(cache): invalidate stale entries
+refactor(pipeline): simplify track dispatch
+```
+
+## Changelog staging
+
+`CHANGELOG_STAGING.md` holds future release notes. Update it once per commit
+session only for release-relevant changes:
+
+- Read only the context needed.
+- Add concise bullets under `Added`, `Changed`, `Fixed`, `Removed`, or `Performance`.
+- Prefer one resulting-change bullet over one per commit; avoid duplicates.
+- Omit routine refactors, formatting, test-only work, and internal details unless
+  release-relevant.

genome_spy_python-0.7.0/.agents/skills/proper-code-review/SKILL.md

@@ -0,0 +1,73 @@
+---
+name: proper-code-review
+description: Review code changes for correctness, simplicity, readability, maintainability, architectural fit, and meaningful system-level performance. Use when asked to review a diff, implementation, branch, or recent code changes.
+---
+
+# Code review
+
+Review as a senior engineer. Prioritize correctness and low lifecycle complexity:
+simple, readable, maintainable code that fits the repository and performs well for
+realistic workloads. Prefer a few high-confidence findings over exhaustive
+commentary; do not manufacture findings.
+
+Added complexity must provide a concrete benefit: correctness, meaningful reuse,
+change locality, real variation, necessary scalability, or material performance.
+Prefer local reasoning and localized changes. Reuse suitable repository concepts,
+avoid duplicated knowledge and parallel mechanisms, and reject speculative
+abstraction/extensibility. Judge performance across the system, not just locally.
+
+## Review method
+
+1. **Understand the change.** Read the diff first. Identify its problem, behavior
+   changes, affected interfaces, invariants, state, and architectural concepts.
+   Read only enough surrounding code to answer concrete questions, not the whole
+   repository. When relevant, inspect callers for compatibility, tests for intended
+   behavior/invariants, existing mechanisms before proposing new ones, analogous
+   implementations for duplication/fit, and broader execution paths for performance.
+2. **Verify hypotheses.** Check concerns against repository evidence when possible.
+   Locate duplicated representations, inspect callers before claiming breakage,
+   search for existing abstractions, and establish performance significance before
+   criticizing or recommending optimization. Do not present assumptions as facts.
+3. **Review correctness.** Look for concrete behavioral bugs/regressions, broken
+   invariants, invalid assumptions, relevant edge cases, unsafe mutations/state
+   transitions, error/resource lifecycle issues, concurrency problems, and
+   unintended compatibility changes. Passing tests are evidence, not proof.
+   Read [correctness](references/correctness.md) for deeper analysis.
+4. **Review design.** Consider simplicity (KISS), readability, maintainability,
+   cohesion, coupling, abstraction, change locality, blast radius, and fit together.
+   Check whether a simpler solution supports local understanding/change, whether
+   synchronized edits duplicate knowledge, whether machinery is unnecessary,
+   whether abstractions share real semantics, and whether scope is proportional.
+   Apply YAGNI: avoid cleverness or future machinery without concrete benefit.
+   Read [design](references/design.md) for deeper analysis.
+5. **Review performance.** Check algorithmic complexity, repeated expensive work,
+   I/O, memory/copying, serialization/data movement, CPU/GPU synchronization, and
+   realistic scaling. Judge end-to-end impact; reject meaningful complexity for
+   negligible local gains. Read [performance](references/performance.md) when relevant.
+6. **Validate findings.** Each problem must be concrete, consequential, supported
+   by evidence, and paired with a proportionate fix that reduces total complexity.
+   Remove speculative, cosmetic, generic, and low-value comments. No finding is
+   better than a weak finding.
+
+## Output
+
+Order findings by practical impact. For each, use:
+
+### `[Severity] Short problem statement`
+
+- **Location:** Smallest useful code region.
+- **Problem:** What is wrong.
+- **Impact:** Why it matters here.
+- **Recommendation:** Smallest reasonable fix.
+
+Severity:
+
+- **Critical:** Correctness, data loss, security, or severe failure.
+- **Important:** Substantial behavioral, architectural, maintainability,
+  performance, or scalability issue.
+- **Minor:** Worthwhile simplification or readability improvement.
+
+Combine findings with the same root cause and state uncertainty explicitly.
+Normally omit stylistic nits. Do not add generic best-practice advice, speculative
+future concerns, praise for unaffected code, or alternative architectures merely
+because they are possible. If none remain, say no significant issues were found.

genome_spy_python-0.7.0/AGENTS.md

@@ -0,0 +1,65 @@
+# Repository instructions
+
+`genome-spy-python` wraps GenomeSpy in an idiomatic Python API, as Altair wraps
+Vega-Lite. Users author compatible specs, serialize JSON, and render in notebooks.
+Treat it as a library with a clean public API. Build the reusable Core grammar
+before app-specific concepts. Notebook rendering is first-class; prefer `anywidget`.
+
+## Layout
+
+| Path | Purpose |
+| --- | --- |
+| `src/genome_spy/` | Installable library |
+| `tests/` | pytest suite |
+| `scripts/` | Helpers run via `uv run` |
+| `plans/` | Plans, context, and implementation notes |
+| `material/` | Local reference documents; not committed |
+| `tmp/` | External reference code/repos; not committed |
+
+## Working rules
+
+- Think before coding: state assumptions and tradeoffs; push back when warranted.
+- Solve the current problem with minimal code. Avoid speculative features and
+  premature abstractions; duplicate a little first. If 200 lines could be 50,
+  rewrite them.
+- Make surgical changes: match existing style, remove imports/variables orphaned
+  by your edits, and leave unrelated code alone.
+- Set verifiable goals and give a brief plan for multi-step work.
+- Keep core logic pure Python; isolate I/O and integration at the edges.
+- Follow the linked [language, docstring, formatting, and gallery standards](docs/contributing/standards.md).
+  Read the relevant sections before Python or gallery edits; they are requirements.
+
+## Testing and handoff
+
+Use pytest. Add or update tests for behavior changes, with fixed seeds for random
+behavior. Keep tests small and focused on one behavior. Apply the "Suspicious
+Check": tighten any test that could pass for the wrong reason.
+
+Use `uv` for environments and dependencies. Before handoff, run targeted tests for
+changed behavior and lint. Format before finalizing non-trivial edits; type-check
+when public Python code changes. Report anything you could not validate locally.
+Ruff is the formatting/lint authority; fix issues rather than suppressing them by
+default.
+
+| Task | Command |
+| --- | --- |
+| Sync environment | `uv sync --dev` |
+| Run tests | `uv run pytest tests/ -x` |
+| Lint | `uv run ruff check .` |
+| Format | `uv run ruff format .` |
+| Type-check | `uv run mypy src/` |
+| Install hooks | `uv run pre-commit install` |
+
+Make small, thematic commits. Use conventional commits:
+`<type>(<scope>): <description>` when a scope helps. Do not rewrite history unless
+explicitly asked.
+
+## Plans and references
+
+| Document | Read when |
+| --- | --- |
+| `plans/implementation_plan.md` | Starting implementation; understand phases and open questions |
+| `plans/deferred.md` | Adding scope; avoid intentionally deferred work |
+| `plans/architecture.md` | Changing public API shape, schema strategy, or notebook integration |
+| `plans/tmp_repo_index.md` | Using `tmp/` repos for wrapper/widget design |
+| `plans/project_bootstrap.md` | Updating or extending the initial scaffold |

{genome_spy_python-0.6.0 → genome_spy_python-0.7.0}/CHANGELOG.md

@@ -5,6 +5,35 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.7.0] - 2026-10-20
+
+### Added
+
+- Typed facet and surround compositions, dendrogram transforms, secondary-offset
+  channels, and the remaining schema and expression additions from GenomeSpy 1.2.0.
+- Discoverable named field arguments on encoding constructors and named setters
+  for inline schema objects, with automatic public exports for generated channels.
+- A DepMap clustered heatmap gallery example adapted from the official GenomeSpy
+  surround specification.
+
+### Changed
+
+- Upgrade the bundled GenomeSpy Core, controls, Inspector, and schema to 1.2.0.
+- Use surround layouts in the PISA interaction matrix, UpSet mutations, PBMC
+  marker matrix, and PBMC cell heatmap examples.
+- Use named Python arguments and schema objects throughout gallery configuration,
+  and Python expression objects in the updated surround examples.
+
+### Fixed
+
+- Keep chart colors aligned with the documentation's selected light or dark theme.
+- Serialize named channel objects in shared surround-side encodings.
+- Preserve secondary-offset field and expression types according to the schema.
+- Support named resolution settings on unit charts and omit forbidden schema
+  properties from generated signatures.
+- Avoid redundant validation of nested schema objects when validation is explicitly
+  disabled, while retaining default validation checks.
+
 ## [0.6.0] - 2026-10-01
 
 ### Added
@@ -152,6 +181,8 @@ First public alpha release, targeting GenomeSpy Core 0.87.0.
 - Packaged example datasets, documentation, tutorials, and an interactive
   visualization gallery.
 
+[0.7.0]: https://github.com/genome-spy/genome-spy-python/compare/v0.6.0...v0.7.0
+[0.6.0]: https://github.com/genome-spy/genome-spy-python/compare/v0.5.0...v0.6.0
 [0.5.0]: https://github.com/genome-spy/genome-spy-python/compare/v0.4.0...v0.5.0
 [0.4.0]: https://github.com/genome-spy/genome-spy-python/compare/v0.3.0...v0.4.0
 [0.3.0]: https://github.com/genome-spy/genome-spy-python/compare/v0.2.0...v0.3.0

{genome_spy_python-0.6.0 → genome_spy_python-0.7.0}/CONTRIBUTING.md

@@ -90,6 +90,7 @@ Regenerate after changing that version or the generator:
 
 ```bash
 uv run python tools/generate_schema_wrapper.py
+uv run python tools/vendor_javascript.py
 ```
 
 This requires `npm` and internet access. Use `--package-dir` for a local package,
@@ -111,6 +112,25 @@ New transforms should be generated from the schema. Add overrides in
 `tools/generate_schema_wrapper.py` only for Python conventions the schema
 cannot describe.
 
+The generator emits explicit typed constructor and setter arguments from schema
+properties, including inline objects; avoid replacing these with untyped
+`**kwargs` or runtime-only signature patches. Channel imports and `__all__` are
+generated too. A new channel must appear in the public namespace automatically.
+New composition families still need renderable chart classes, serialization and
+round-trip support. The schema-derived root coverage test must fail until that
+support exists, rather than silently omitting unfamiliar roots.
+
+When adapting examples, use named schema objects or nested setters such as
+`gs.Scale().zoom(extent="data")`. Keep dictionaries for data records and authored
+names such as datasets, templates and parameter values. Test generated signatures
+and type-check a consumer example as well as validating its serialized JSON.
+
+Altair uses the same separation between a generated schema API and handwritten
+chart operations. Its [constructor and setter generator](https://github.com/vega/altair/blob/main/tools/schemapi/codegen.py)
+emits typed signatures and overloads; its [property-setter runtime](https://github.com/vega/altair/blob/main/altair/utils/schemapi.py)
+also supplies signatures for interactive completion. Prefer our existing explicit
+Python methods so completion, static checking and runtime calls agree.
+
 ## Documentation
 
 ### API reference

{genome_spy_python-0.6.0 → genome_spy_python-0.7.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: genome-spy-python
-Version: 0.6.0
+Version: 0.7.0
 Summary: genome-spy-python is a declarative genomics visualization library for Python, built on top of the genome-spy JSON specification.
 Project-URL: Homepage, https://github.com/genome-spy/genome-spy-python
 Project-URL: Documentation, https://genomespy.app/genome-spy-python/
@@ -33,6 +33,8 @@ Requires-Python: >=3.11
 Requires-Dist: anywidget>=0.11.0
 Requires-Dist: jsonschema>=4.26.0
 Requires-Dist: traitlets>=5.14.3
+Provides-Extra: anndata
+Requires-Dist: anndata>=0.11.0; extra == 'anndata'
 Provides-Extra: arrow
 Requires-Dist: pyarrow>=18.0.0; extra == 'arrow'
 Description-Content-Type: text/markdown
@@ -100,6 +102,21 @@ for supported tables and live updates.
 
 See the [getting-started guide](docs/getting-started.md) for the first example.
 
+## Agent skill
+
+The package bundles a visualization skill for Codex and Claude Code. Installation
+is explicit; importing or installing the Python package does not modify agent
+configuration. To install for both agents in the current project:
+
+```bash
+genome-spy-install-skills
+```
+
+Use `--agent codex` or `--agent claude` to select one agent, `--project PATH` for
+another project, or `--user` for personal installation. Existing modified skills
+are preserved unless `--force` is supplied. See
+[installation and update details](docs/agents/skill-distribution.md).
+
 ## Examples
 
 A zoomable Manhattan plot using the bundled HapMap data (requires `pandas`):

{genome_spy_python-0.6.0 → genome_spy_python-0.7.0}/README.md

@@ -61,6 +61,21 @@ for supported tables and live updates.
 
 See the [getting-started guide](docs/getting-started.md) for the first example.
 
+## Agent skill
+
+The package bundles a visualization skill for Codex and Claude Code. Installation
+is explicit; importing or installing the Python package does not modify agent
+configuration. To install for both agents in the current project:
+
+```bash
+genome-spy-install-skills
+```
+
+Use `--agent codex` or `--agent claude` to select one agent, `--project PATH` for
+another project, or `--user` for personal installation. Existing modified skills
+are preserved unless `--force` is supplied. See
+[installation and update details](docs/agents/skill-distribution.md).
+
 ## Examples
 
 A zoomable Manhattan plot using the bundled HapMap data (requires `pandas`):

{genome_spy_python-0.6.0 → genome_spy_python-0.7.0}/THIRD_PARTY_NOTICES.md

@@ -106,6 +106,19 @@ The full CC BY 4.0 and CC0 texts are in `LICENSES/CC-BY-4.0.txt` and
 `LICENSES/DATA-SOURCE-REPOSITORIES-MIT.txt` accompany the corresponding
 selection, formatting and example contributions.
 
+## PBMC marker expression
+
+`pbmc_markers.json.gz` contains selected markers from 2,638 cells in Scanpy's processed
+PBMC3k fixture, obtained from [cellxgene revision
+68dfbcc](https://github.com/chanzuckerberg/cellxgene/blob/68dfbcc2eb675e96c6a5e2a6b7a0d3465ccf46bc/example-dataset/pbmc3k.h5ad).
+The original [10x Genomics PBMC3k dataset](https://www.10xgenomics.com/datasets/3-k-pbm-cs-from-a-healthy-donor-1-standard-1-1-0)
+is licensed CC BY 4.0; credit 10x Genomics and Scanpy's preprocessing contributors.
+The subset retains saved UMAP coordinates, `raw.X` log1p counts and cell-type annotations, groups
+cells while preserving within-group order, and adds marker means, a single-linkage
+Euclidean group tree, and plotting coordinates. `tools/prepare_pbmc_gallery_data.py` reproduces these modifications;
+the source checksum is retained in the bundle. The CC BY 4.0 text is in
+`LICENSES/CC-BY-4.0.txt`.
+
 ## Bundled TCGA example tables
 
 | Files | Source and processing |

genome_spy_python-0.7.0/docs/_ext/genomespy_agents.py

@@ -0,0 +1,36 @@
+"""Publish plain agent Markdown after successful HTML builds."""
+
+from pathlib import Path
+import sys
+from typing import Any
+
+from sphinx.application import Sphinx
+
+_ROOT = Path(__file__).resolve().parents[2]
+sys.path.insert(0, str(_ROOT))
+
+from tools.agent_docs import publish_agent_docs  # noqa: E402
+
+
+def _publish(app: Sphinx, exception: Exception | None) -> None:
+    if exception is None and app.builder.format == "html":
+        publish_agent_docs(_ROOT, Path(app.outdir), app.config.html_baseurl)
+
+
+def setup(app: Sphinx) -> dict[str, Any]:
+    """Register agent-doc publication after successful HTML builds.
+
+    Args:
+        app: Sphinx application.
+
+    Returns:
+        Extension version and parallel-read safety metadata.
+
+    Raises:
+        None.
+
+    Example:
+        >>> setup(app)  # doctest: +SKIP
+    """
+    app.connect("build-finished", _publish)
+    return {"version": "1", "parallel_read_safe": True}

{genome_spy_python-0.6.0 → genome_spy_python-0.7.0}/docs/_ext/genomespy_gallery.py

@@ -107,6 +107,15 @@ def _remove_stale_arrow_assets(referenced: set[str]) -> None:
 # Cards share titles, but only the gallery index shows data subtitles.
 # Example headings stay unchanged; new examples fall back to their source title.
 GALLERY_CARD_LABELS = {
+    "pbmc_umap": ("PBMC UMAP", "PBMC3k · Cell types"),
+    "pbmc_umap_genes": ("PBMC UMAP panels", "PBMC3k · Markers and cell types"),
+    "pbmc_marker_matrix": ("PBMC marker matrix", "PBMC3k · 12 marker genes"),
+    "pbmc_cell_heatmap": ("PBMC cell heatmap", "PBMC3k · 2,638 cells"),
+    "pbmc_expression_tracks": ("PBMC expression tracks", "PBMC3k · AnnData"),
+    "clustered_heatmap": (
+        "Clustered heatmap with dendrograms",
+        "DepMap · 60 cell lines and 72 genes",
+    ),
     "stacked_bar": ("Horizontal stacked bars", "Barley yields · Vega datasets"),
     "layered_lollipop": ("Layered lollipop plot", "Synthetic sine wave"),
     "independent_scales": (

{genome_spy_python-0.6.0 → genome_spy_python-0.7.0}/docs/_static/genomespy.css

@@ -432,6 +432,7 @@ p.lead {
 /* ---------- Live documentation embeds ---------- */
 /* Render gallery and prose charts inline at the full available width. */
 .gs-doc-embed {
+  color-scheme: light;
   display: block;
   width: 100%;
   height: 400px;
@@ -439,6 +440,17 @@ p.lead {
   margin: 1.25rem 0 0.5rem;
 }
 
+/* Match Furo's selected theme instead of the operating system's preference. */
+body[data-theme="dark"] .gs-doc-embed {
+  color-scheme: dark;
+}
+
+@media (prefers-color-scheme: dark) {
+  body:not([data-theme="light"]) .gs-doc-embed {
+    color-scheme: dark;
+  }
+}
+
 .gs-embed-actions {
   margin: 0.3rem 0 0.6rem;
   text-align: center;

genome_spy_python-0.7.0/docs/agents/data.md

@@ -0,0 +1,121 @@
+# Data inputs
+
+## Records: base package
+
+Use when the input is a list of records with numeric `x` and `y`, plus an `id`.
+
+```python
+import genome_spy as gs
+
+rows = [{"id": "a", "x": 2, "y": 5}, {"id": "b", "x": 4, "y": 3}]
+chart = gs.Chart(rows).mark_point().encode(x="x:Q", y="y:Q", tooltip=["id:N"])
+chart.to_dict()
+```
+
+## Dataframes and indexes: `arrow` extra
+
+Pandas, Polars, and PyArrow tables are accepted directly. Expose useful indexes
+as columns; inspect types, bounded samples, missing values, ranges, and order.
+
+```python
+import genome_spy as gs
+import pandas as pd
+
+rows = [{"id": "a", "x": 2, "y": 5}, {"id": "b", "x": 4, "y": 3}]
+frame = pd.DataFrame(rows).set_index("id")
+records_with_ids = frame.reset_index()
+assert records_with_ids["id"].is_unique
+chart = gs.Chart(records_with_ids).mark_point().encode(
+    x="x:Q", y="y:Q", tooltip=["id:N"]
+)
+chart.to_dict()
+```
+
+CSV loading and data checks happen in Python; [declarative transforms](encodings.md#browser-transforms)
+execute in GenomeSpy. Preserve supplied measurements and report any filtering,
+sampling, conversions, or derived columns.
+
+## Measurement source and selection scope
+
+Before encoding, identify the source table/field, units, transformations and
+aggregation. Normalizing before grouping differs from normalizing group means;
+fractions also need a denominator and missing/zero rules.
+
+Feature selection and record filtering are separate. Here a supplied feature
+list chooses columns to show; both observation groups remain visible. The means
+are prepared inputs, not calculated by this chart.
+
+```python
+import genome_spy as gs
+
+summaries = [
+    {"group": "A", "feature": "f1", "mean_normalized_score": 0.2},
+    {"group": "B", "feature": "f1", "mean_normalized_score": 0.8},
+    {"group": "A", "feature": "f2", "mean_normalized_score": 0.5},
+    {"group": "B", "feature": "f2", "mean_normalized_score": 0.4},
+]
+feature_lists = {"A": ["f1"], "B": ["f2"]}
+selected_features = feature_lists["A"]
+visible_groups = ["A", "B"]
+selected = [
+    row for row in summaries
+    if row["feature"] in selected_features and row["group"] in visible_groups
+]
+chart = gs.Chart(selected).mark_point(filled=True).encode(
+    x=gs.X("feature:N").scale(type="band", domain=selected_features),
+    y=gs.Y("group:N").scale(type="band", domain=visible_groups, reverse=True),
+    color=gs.Color("mean_normalized_score:Q").legend(title="Mean normalized score"),
+)
+chart.to_dict()
+```
+
+Expected: two points for `f1`, one per group. Changing `visible_groups` filters
+records; changing its order changes display order. Choosing a different value
+field changes the measurement. Keep these decisions explicit.
+
+## Remote CSV
+
+```python
+import genome_spy as gs
+
+remote = gs.Chart(
+    gs.Data(url="https://example.org/measurements.csv", format=gs.data_format(type="csv"))
+).mark_point().encode(x="x:Q", y="y:Q")
+remote.to_dict()  # Serializes the URL; does not fetch or validate its contents.
+```
+
+Replace the illustrative URL with an accessible table. `inline=True` bundles the
+runtime, not URL data; rendering still needs network access.
+
+## AnnData projection: `anndata` and `arrow` extras
+
+A small synthetic input makes the source selection explicit:
+
+```python
+import anndata as ad
+import genome_spy as gs
+import numpy as np
+import pandas as pd
+from genome_spy.anndata import from_anndata
+
+adata = ad.AnnData(
+    X=np.array([[1.0], [2.0]]),
+    obs=pd.DataFrame({"group": ["A", "B"]}, index=["a", "b"]),
+    var=pd.DataFrame(index=["feature_a"]),
+)
+adata.obsm["embedding"] = np.array([[0.0, 1.0], [2.0, 3.0]])
+snapshot = from_anndata(
+    adata, obs=["group"], genes=["feature_a"], obsm_keys=[("embedding", 0)]
+).reset_index()
+projected = gs.Chart(snapshot).mark_point().encode(
+    x="embedding-0:Q", y="feature_a:Q", color="group:N"
+)
+projected.to_dict()
+```
+
+`from_anndata` reads selected `X` genes, `obs` fields, and zero-based `obsm`
+components into a pandas snapshot. It does not select `.raw` or a layer.
+Resolve the intended source before calling it; no analysis is performed.
+Expected: two points with `embedding-0` on x, `feature_a` on y, colored by group.
+
+Continue with [encodings](encodings.md) or [quickstart](quickstart.md).