feat(altair): implement manhattan-gwas

[github-actions]

Implementation: manhattan-gwas - python/altair

Implements the python/altair version of manhattan-gwas.

File: plots/manhattan-gwas/implementations/python/altair.py

Parent Issue: #2925

---

🤖 impl-generate workflow

[claude]

AI Review - Attempt 1/3

Image Description

Light render (plot-light.png): The Manhattan plot displays on a warm off-white background (#FAF8F1). The title "Manhattan Plot: GWAS Results" is positioned at the top in dark text. SNPs are scattered across chromosomes (1-22) with alternating Okabe-Ito colors: green (#009E73), orange (#D55E00), blue (#0072B2), pink (#CC79A7), and yellow (#E69F00). Two threshold lines are visible: a dashed line at y≈7.3 (genome-wide significance) and a lighter dashed line at y≈5 (suggestive threshold). Chromosome labels are positioned below. X-axis shows "Genomic Position", y-axis shows "-log₁₀(p-value)". All text is clearly readable with properly sized title (28px), axis labels (22px), and tick labels (18px). Data points above threshold are slightly larger (size=100) than background points (size=60). Legibility verdict: PASS — all text readable, colors distinct from light background, grid lines subtle.

Dark render (plot-dark.png): The same plot rendered on a warm dark background (#1A1A17) with light-colored text. Title and axis labels are in light colors (#F0EFE8) for proper contrast. Data colors are identical to light render — all chromosomes maintain their respective Okabe-Ito colors (green, orange, blue, pink, yellow). Grid lines remain subtle at opacity 0.10. Threshold lines render with proper contrast. Chromosome labels are clearly visible in light text. Legibility verdict: PASS — theme-correct chrome, readable text, no dark-on-dark failures in either render.

Score: 87/100

Category	Score	Max
Visual Quality	30	30
Design Excellence	10	20
Spec Compliance	14	15
Data Quality	15	15
Code Quality	10	10
Library Mastery	8	10
Total	87	100

Visual Quality (30/30)

• VQ-01: Text Legibility (8/8) - All font sizes explicitly set; title 28px, labels 22px, ticks 18px. Perfectly readable in both themes.
• VQ-02: No Overlap (6/6) - Chromosome labels well-spaced, no overlapping text elements
• VQ-03: Element Visibility (6/6) - Opacity 0.7 adapts optimally to 100k+ SNPs; size condition (100 vs 60) emphasizes peaks
• VQ-04: Color Accessibility (2/2) - Okabe-Ito palette is CVD-safe; good contrast throughout
• VQ-05: Layout & Canvas (4/4) - Plot fills canvas optimally with balanced margins
• VQ-06: Axis Labels & Title (2/2) - Descriptive labels with notation
• VQ-07: Palette Compliance (2/2) - First series #009E73 ✓; alternating Okabe-Ito colors ✓; backgrounds #FAF8F1/#1A1A17 ✓; theme chrome correct in both ✓

Design Excellence (10/20)

• DE-01: Aesthetic Sophistication (4/8) - Well-configured defaults; thoughtful theme application but lacks distinctive design innovation
• DE-02: Visual Refinement (4/6) - Custom styling on threshold lines, subtle grid (0.10 opacity), refined text positioning
• DE-03: Data Storytelling (2/6) - Data displayed clearly; size variation emphasizes peaks but lacks additional visual hierarchy

Spec Compliance (14/15)

• SC-01: Plot Type (5/5) - Correct Manhattan plot with all required features
• SC-02: Required Features (4/4) - Alternating colors, thresholds, cumulative positions, chromosome labels, size variation all present
• SC-03: Data Mapping (3/3) - X/Y correctly assigned; all data visible
• SC-04: Title & Legend (2/3) - Title should follow format: manhattan-gwas · altair · anyplot.ai instead of "Manhattan Plot: GWAS Results"

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6) - Shows non-significant, suggestive, and genome-wide significant peaks with varied chromosome patterns
• DQ-02: Realistic Context (5/5) - Scientifically plausible GWAS simulation with neutral context
• DQ-03: Appropriate Scale (4/4) - Correct p-value ranges, chromosome proportions, and threshold standards (5×10⁻⁸, 1×10⁻⁵)

Code Quality (10/10)

• CQ-01: KISS Structure (3/3) - Linear structure; no unnecessary functions or classes
• CQ-02: Reproducibility (2/2) - np.random.seed(42) set for deterministic output
• CQ-03: Clean Imports (2/2) - Only necessary packages imported
• CQ-04: Code Elegance (2/2) - Clean, Pythonic, appropriate complexity, no fake functionality
• CQ-05: Output & API (1/1) - Correct output format

Library Mastery (8/10)

• LM-01: Idiomatic Usage (5/5) - Expert altair patterns: mark_circle, encode, layer, vconcat, conditional sizing
• LM-02: Distinctive Features (3/5) - Uses conditional encoding and layering; could leverage more advanced altair capabilities

Score Caps Applied

• None / All criteria within normal ranges

Strengths

• Perfect visual quality with excellent text legibility and theme-correct palette compliance in both renders
• Excellent data quality with realistic GWAS simulation using correct thresholds and chromosome proportions
• Perfect code quality with clean structure, reproducibility, and idiomatic altair usage
• Strong theme adaptation across both renders with proper chrome colors and legibility
• All technical spec requirements implemented correctly with proper threshold visualization

Weaknesses

• Title format does not match standard: should be manhattan-gwas · altair · anyplot.ai instead of "Manhattan Plot: GWAS Results"
• Design Excellence is low (10/20): lacks aesthetic sophistication beyond library defaults
• Visual storytelling is minimal (DE-03=2): no visual hierarchy or emphasis beyond basic size variation

Issues Found

1. SC-04 MINOR: Title format — should be {spec-id} · {library} · anyplot.ai
   • Fix: Change title to "manhattan-gwas · altair · anyplot.ai"
2. DE-01 LOW: Aesthetic sophistication — relies on library defaults
   • Fix: Apply custom color harmony or refined styling to elevate design
3. DE-03 LOW: Data storytelling — no visual hierarchy beyond size
   • Fix: Add color emphasis or visual hierarchy to highlight significant associations

AI Feedback for Next Attempt

Strong technical foundation with perfect visual quality and palette compliance. To reach publication quality (≥90), improve design excellence by: (1) fixing title format to manhattan-gwas · altair · anyplot.ai, (2) enhancing aesthetic sophistication through refined styling or custom accent colors, (3) strengthening data storytelling by using color contrast or visual emphasis to highlight peaks above the genome-wide significance threshold. The threshold lines are well-styled; consider accent coloring those peaks or subtle background highlights to create stronger visual focal points.

Verdict: REJECTED

Score 87/100 is below the Attempt 1 threshold of 90/100. The implementation is technically excellent but needs design enhancements to reach publication quality. Focus repairs on title format, aesthetic sophistication, and visual hierarchy for significant peaks.

[github-actions]

🔧 Repair Attempt 1/4

Applied fixes based on AI review feedback.

Status: Repair completed, re-triggering review...

---

🤖 impl-repair

[claude]

AI Review - Attempt 2/3

Image Description

Light render (plot-light.png): The plot displays on a warm off-white background (#FAF8F1) with excellent clarity. The title "Manhattan Plot: GWAS Results" and subtitle are prominently displayed in dark text. The main scatter plot shows thousands of SNP points arranged by cumulative genomic position, with alternating Okabe-Ito colors across chromosomes (starting with #009E73 green in the first chromosome). Two dashed threshold lines are visible at -log10(p) = 7.3 (genome-wide significance) and 5 (suggestive). The y-axis label "-log₁₀(p-value)" and chromosome labels (1-22) at the bottom are all clearly readable against the light background. Point sizes vary based on significance, creating visual emphasis on peaks. Grid lines are subtle at 10% opacity. All text is readable and properly colored using INK (#1A1A17) for titles and INK_SOFT (#4A4A44) for secondary elements.

Dark render (plot-dark.png): The plot transitions seamlessly to the warm near-black background (#1A1A17) with all chrome elements properly inverted. The title, axis labels, and tick labels are now rendered in light text (INK: #F0EFE8, INK_SOFT: #B8B7B0), maintaining excellent legibility. Importantly, the data colors (Okabe-Ito palette positions 1-6) remain identical to the light render—only the chrome has flipped. The threshold lines are still visible, and the grid is appropriately subtle. No "dark-on-dark" failures are present; all text contrasts properly against the dark background. The plot is fully readable and visually balanced on both surfaces.

Score: 88/100

Category	Score	Max
Visual Quality	30	30
Design Excellence	11	20
Spec Compliance	14	15
Data Quality	15	15
Code Quality	10	10
Library Mastery	8	10
Total	88	100

Visual Quality (30/30)

• VQ-01: Text Legibility (8/8) — Title at 28px, axis labels at 22px, tick labels at 18px; all readable in both themes with proper INK tokens
• VQ-02: No Overlap (6/6) — Clean layout with no collisions; chromosome labels positioned below with spacing
• VQ-03: Element Visibility (6/6) — Points clearly visible with opacity=0.7 and size variation (60–100px); no occlusion issues
• VQ-04: Color Accessibility (2/2) — Okabe-Ito palette (CVD-safe); first series is #009E73; adequate contrast
• VQ-05: Layout & Canvas (4/4) — Good proportions; generous spacing; nothing cut off
• VQ-06: Axis Labels & Title (2/2) — Descriptive title and subtitle; axis labels clear
• VQ-07: Palette Compliance (2/2) — First series #009E73 ✓; alternating chromosomes follow Okabe-Ito; backgrounds correct (#FAF8F1 light, #1A1A17 dark); both renders theme-compliant

Design Excellence (11/20)

• DE-01: Aesthetic Sophistication (4/8) — Functional design using standard Okabe-Ito without additional refinement; could benefit from custom styling or emphasis techniques
• DE-02: Visual Refinement (3/6) — Clean grid and threshold lines; spines handled subtly; could improve with refined line weights or additional visual polish
• DE-03: Data Storytelling (4/6) — Size variation creates hierarchy (significant points larger); threshold lines guide focus; could strengthen with annotations on peaks or more explicit emphasis

Spec Compliance (14/15)

• SC-01: Plot Type (5/5) — Correct Manhattan plot; cumulative genomic position on x-axis; -log10(p-value) on y-axis; alternating chromosome colors
• SC-02: Required Features (4/4) — Alternating colors ✓; threshold line at 7.3 ✓; suggestive threshold at 5 ✓; point size reduction for density ✓
• SC-03: Data Mapping (3/3) — X-axis correctly shows cumulative genomic position; y-axis correctly shows -log10 transformed p-values; all data visible
• SC-04: Title & Legend (2/3) — Title "Manhattan Plot: GWAS Results" is descriptive but missing required format (manhattan-gwas · altair · anyplot.ai); legend appropriately omitted for this plot type

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6) — Realistic GWAS dataset with baseline noise, suggestive hits, and significant peaks across multiple chromosomes
• DQ-02: Realistic Context (5/5) — Simulated data is plausible; realistic chromosome lengths; appropriate SNP density; neutral content
• DQ-03: Appropriate Scale (4/4) — Y-axis scale sensible for p-value magnitudes; x-axis spans full genome; thresholds at meaningful values

Code Quality (10/10)

• CQ-01: KISS Structure (3/3) — No unnecessary functions; straightforward linear structure
• CQ-02: Reproducibility (2/2) — np.random.seed(42) ensures deterministic output
• CQ-03: Clean Imports (2/2) — Only necessary imports (os, altair, numpy, pandas); all used
• CQ-04: Code Elegance (2/2) — Appropriate complexity; no fake functionality
• CQ-05: Output & API (1/1) — Saves as plot-{THEME}.png and plot-{THEME}.html ✓; current API

Library Mastery (8/10)

• LM-01: Idiomatic Usage (5/5) — Proper high-level API; layer composition (alt.layer()); conditional encoding (alt.condition() for size); theme-adaptive configuration
• LM-02: Distinctive Features (3/5) — Layer composition and tooltips are well-executed; could leverage more Altair-specific features (transforms, selection/filtering)

Score Caps Applied

• None

Strengths

• Excellent legibility in both light and dark themes with proper theme token usage
• Realistic, well-structured GWAS data generation with appropriate chromosome distribution
• Clean, reproducible code with deterministic seeding
• Proper use of Altair's layer composition and conditional encoding
• Correct Okabe-Ito palette application with alternating chromosome colors
• Threshold lines effectively highlight significance levels
• Interactive tooltips enhance usability

Weaknesses

• Title format deviates from required spec-id/library/anyplot.ai format
• Design excellence could be higher with additional visual refinement or emphasis on key findings
• Limited use of Altair's advanced features beyond basic layer composition

Issues Found

1. SC-04 MINOR: Title format missing required elements

   • Fix: Change title to include spec-id and library designation (e.g., "manhattan-gwas · altair · anyplot.ai")

2. DE-01/DE-03 MODERATE: Design could be more visually sophisticated

   • Fix: Add annotations for the top peaks, use color contrast to emphasize significant regions, or refine line weights for visual hierarchy

AI Feedback for Next Attempt

The implementation is solid and publication-ready. For the next iteration, focus on: (1) updating the title format to include spec-id and library, (2) enhancing visual storytelling by annotating the most significant peaks or using accent colors for emphasis, and (3) refining line weights and grid styling for greater sophistication.

Verdict: APPROVED