Practice Lab (Interactive)
Hands-on mini exercises. Adjust parameters and watch how summaries change—this builds intuition for QC, trimming, and filtering decisions. All data here is synthetic and runs in your browser.
Exercise 1 — Phred score ↔ error probability
Convert between Phred score $Q$ and error probability $p$ using $Q=-10\log_{10}(p)$.
Typical good Illumina bases are around Q30+
Q30 corresponds to p=0.001
Exercise 2 — Trimming tradeoffs (quality cutoff + minimum length)
Raising the quality cutoff keeps fewer bases; raising minimum length drops short reads entirely. The best thresholds depend on the downstream task.
Common starting points: Q15–Q25
Dropping too short reads can reduce false alignments
Exercise 3 — Variant filtering (DP, GQ, VAF)
Filtering changes both sensitivity and precision. Slide thresholds and see how many variants remain.
Depth too low increases false positives and genotype uncertainty
Genotype quality is caller-specific but useful as a coarse filter
For germline diploid, hets cluster around ~0.5 (but coverage/bias matter)
Next ideas (if you want even more interactivity)
- Interactive PCA/UMAP explorer (choose normalization & outlier handling)
- Alignment parameter sandbox (seed length, mismatch penalty) and mapping outcomes
- Contamination mixture simulator and GC/read-length changes
- Single-cell cell filtering (nUMI/percent.mt thresholds) with live plots