skills/scientific-writing/scientific-slides/SKILL.md
Scientific presentations for conferences, seminars, thesis defenses, and grant pitches. Slide design, talk structure, timing, data viz for slides, QA. PowerPoint and LaTeX Beamer. For posters use latex-research-posters.
npx skillsauth add jaechang-hits/sciagent-skills scientific-slidesInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Scientific presentations are a critical medium for communicating research at conferences, seminars, defenses, and professional talks. This knowhow covers end-to-end presentation development: structure and content planning, visual design principles, data visualization adaptation, timing and pacing, and quality assurance across PowerPoint and LaTeX Beamer formats.
| Talk Type | Duration | Slides | Focus | Key Finding Count | |-----------|----------|--------|-------|-------------------| | Conference talk | 10–20 min | 12–20 | 1–2 key findings | 1–2 | | Academic seminar | 45–60 min | 40–60 | Comprehensive coverage | 3–6 | | Thesis defense | 45–60 min | 45–65 | Full dissertation | All studies | | Grant pitch | 10–20 min | 12–18 | Significance + feasibility | Preliminary data | | Journal club | 20–45 min | 20–40 | Critical analysis | Paper's findings |
Visual-first approach: Start with visuals (figures, diagrams, images), then add text as support. Target 60–70% visual content, 30–40% text. Every slide should have a strong visual element.
Typography:
Color:
Layout:
Key differences from journal figures:
| Chart Type | Best For | Slide Adaptation | |-----------|----------|-----------------| | Bar chart | Category comparison | Max 6–8 bars, large labels | | Line graph | Trends over time | Bold lines, 2–3 series max | | Scatter plot | Correlations | Large points, trend line | | Heatmap | Matrix patterns | High contrast, annotate key cells | | Flowchart | Methodology | Build step-by-step with animations |
Every scientific talk follows this narrative structure:
Start: What is your priority?
├── Mathematical content, equations, version control?
│ └── YES → LaTeX Beamer (see assets/beamer templates)
├── Editable slides, company templates, animations?
│ └── YES → PowerPoint (programmatic or template-based)
├── Fast creation, non-technical audience, visual impact?
│ └── YES → PowerPoint or image-based PDF
└── Not sure
└── PowerPoint (most flexible default)
| Duration | Simple Topic | Average | Complex Topic | |----------|-------------|---------|---------------| | 5 min | 5–6 | 6–8 | 5–7 | | 10 min | 10–12 | 12–14 | 10–12 | | 15 min | 14–16 | 16–18 | 14–16 | | 30 min | 25–30 | 30–35 | 25–30 | | 45 min | 38–45 | 45–50 | 38–45 | | 60 min | 50–55 | 55–65 | 50–60 |
General rule: ~1 slide per minute. Complex slides (results, methodology) may take 2–3 minutes; simple slides (transitions, section dividers) take 15–30 seconds.
| Section | % of Time | 15-min Talk | 45-min Talk | |---------|-----------|------------|------------| | Introduction | 15–20% | 2–3 min | 7–9 min | | Methods | 15–20% | 2–3 min | 7–9 min | | Results | 40–50% | 6–7 min | 18–22 min | | Discussion | 15–20% | 2–3 min | 7–9 min | | Conclusion | 5% | 45 sec | 2 min |
MANDATORY: Every slide must have a strong visual element — figure, chart, diagram, image, or icon. Text-only bullet list slides fail to communicate science effectively. Target minimum 2 visual elements per content slide.
MANDATORY: Practice with a timer at least 3 times before presenting. Set timing checkpoints: for a 15-minute talk, check at 3–4 min (finishing intro), 7–8 min (midway through results), 12–13 min (starting conclusions).
Use minimal text as visual support. 3–4 bullets per slide, 4–6 words per bullet. Text is the supporting role; visuals are the stars. Never put full paragraphs on slides.
Include proper citations. Cite 3–5 papers in the introduction (establishing context) and 3–5 in the discussion (comparison). Use author-year format (Smith et al., 2023) for readability.
Design section dividers with visual breaks. Insert visually distinctive slides between major sections (intro → methods → results → conclusion). These help the audience reset and follow the narrative.
Anti-pattern — using default templates without customization. Default PowerPoint/Beamer themes signal "minimal effort." Choose a modern color palette, customize fonts, and add visual personality matching your topic.
Simplify journal figures for slides. Increase all labels to 18–24 pt, remove non-essential panels, use direct labeling instead of legends, emphasize the key finding with color or annotation.
Anti-pattern — cramming full paper content into slides. A 15-minute talk should cover 1–2 key findings, not the entire paper. Leave details for the written paper and prepare backup slides for Q&A.
Prepare backup slides. Put additional data, detailed methods, alternative analyses after the "Thank You" slide. Reference them during Q&A without disrupting the main talk flow.
Never skip conclusions. If running behind, cut earlier content (skip a results slide or compress methods). The conclusion is the audience's take-away message — skipping it wastes the entire talk.
Text-heavy, visual-poor slides. Walls of text, no images or graphics, bullet points as the only content. How to avoid: Start slide creation with visuals first (which figure/diagram?), then add minimal text as support.
Font sizes too small (under 24pt body text). Back-row audience can't read, slides look cramped. How to avoid: Set body text to 24–28 pt, titles to 36–44 pt. Test by viewing slides at 50% zoom — if you can't read it, the audience can't either.
Too many findings for the time slot. Trying to present 5 findings in a 15-minute talk rushes everything. How to avoid: Conference talks = 1–2 findings. Seminars = 3–5. Choose ruthlessly.
Missing research context (no citations). Claims without supporting literature undermine credibility. How to avoid: Search literature before creating slides. Cite 3–5 papers in intro and 3–5 in discussion.
Inconsistent formatting across slides. Different fonts, colors, and layouts from slide to slide look unprofessional. How to avoid: Use master slides/templates. Define your color palette, fonts, and layout grid before starting content.
Low contrast text on background. Light gray text on white, or colored text on busy images. How to avoid: Ensure text-background contrast ratio ≥ 7:1. Use solid color overlays on image backgrounds.
Not practicing with timer. First run-through is during the actual presentation, causing time overruns. How to avoid: Practice minimum 3 times. Mark timing checkpoints on your notes.
Skipping conclusions when running over time. Rushing through or omitting the take-away message. How to avoid: Cut earlier content (a results slide) rather than the conclusion. Prepare a "Plan B" with marked skip-able slides.
# Basic compilation
pdflatex presentation.tex
# With bibliography
pdflatex presentation.tex && bibtex presentation && pdflatex presentation.tex && pdflatex presentation.tex
# Better font support
lualatex presentation.tex
Detailed reference files in references/:
| File | Content |
|------|---------|
| talk_types_guide.md | Detailed structure and strategy for each talk type (conference, seminar, defense, grant pitch, journal club) with example outlines |
| slide_design_guide.md | Extended design principles: color theory, typography tables, layout patterns, accessibility guidelines, Gestalt principles for visual composition |
Templates and guides in assets/:
| File | Description |
|------|-------------|
| beamer_template_conference.tex | LaTeX Beamer template for 15-minute conference talks |
| beamer_template_seminar.tex | LaTeX Beamer template for 45-minute academic seminars |
| beamer_template_defense.tex | LaTeX Beamer template for thesis/dissertation defenses |
| timing_guidelines.md | Comprehensive timing and pacing strategies for all talk types |
development
Bulk RNA-seq DE with R/Bioconductor DESeq2. Negative binomial GLM, empirical Bayes shrinkage, Wald/LRT tests, multi-factor designs, Salmon tximeta import, apeglm LFC shrinkage, MA/volcano/heatmap viz. R gold standard. Use pydeseq2-differential-expression for Python; use edgeR for TMM normalization.
tools
Guide to quality filtering raw VCF files before computing summary stats (Ts/Tv ratio, variant counts, AF distributions). Covers detecting raw VCFs via FILTER column and QUAL inspection, QUAL-based filtering with bcftools, Ts/Tv interpretation, and when NOT to filter. Read before any variant-level QC task. See bcftools-variant-manipulation for advanced filters, gatk-variant-calling for caller config, samtools-bam-processing for upstream alignment QC.
tools
Annotate and filter VCF variants with SnpEff and SnpSift. SnpEff predicts functional effects (HIGH/MODERATE/LOW/MODIFIER), genes, transcripts, AA changes, HGVS; SnpSift filters and adds ClinVar/dbSNP. Java CLI with Python subprocess integration. Use ANNOVAR for multi-database annotation; Ensembl VEP for REST API; SnpEff for fast CLI with pre-built genomes.
tools
GWAS and population genetics tool. Processes PLINK (.bed/.bim/.fam), VCF, and BGEN; runs QC (MAF, HWE, missingness), IBD estimation, PCA, and linear/logistic regression GWAS. Outputs Manhattan-ready summary stats. Use regenie or SAIGE for biobanks (>100k samples) needing mixed models.