從原始數據到生物學洞見——12 個步驟帶你掌握單細胞轉錄組分析的核心流程，搭配互動模擬與程式碼範例。
From raw data to biological insights — master single-cell transcriptomics in 12 steps, with interactive simulations and code examples.
根據 nFeature、nCount、MT% 過濾低品質細胞，移除空液滴與 doublets。
Filter low-quality cells by nFeature, nCount, MT%. Remove empty droplets and doublets.
校正細胞間定序深度差異，比較 LogNormalize、Scran 與 SCTransform。
Correct sequencing depth differences. Compare LogNormalize, Scran, and SCTransform.
挑選具有高生物變異的基因，作為 PCA 與聚類的核心輸入特徵。
Select genes with highest biological variation as key input features for PCA and clustering.
Z-score 標準化讓基因在相近尺度下進入下游分析。
Z-score normalization gives every gene equal weight in downstream analysis.
壓縮基因維度，保留關鍵變異、降低噪音。
Compress gene dimensions — retain key variation, reduce noise.
在 PCA 空間中建立近鄰圖，以 Louvain/Leiden 辨識細胞族群。
Build neighbor graphs in PCA space. Identify cell populations via Louvain/Leiden.
將高維數據投影到 2D，觀察細胞群間的相對關係。
Project high-dimensional data to 2D to visualize cell population relationships.
利用 marker genes 與參考數據集，為 cluster 指定細胞類型名稱。
Assign cell type identities to clusters using marker genes and reference datasets.
比較不同 cluster 或 condition 間的基因表達差異。
Compare gene expression differences between clusters or conditions.
整合多樣本/多批次資料，降低 batch effect 並保留真實生物差異。
Integrate multi-sample/multi-batch data. Remove batch effects while preserving true biology.
分析 ligand-receptor 交互作用，探索細胞間訊號傳遞與微環境調控。
Analyze ligand-receptor interactions. Explore intercellular signaling and microenvironment regulation.
推估細胞狀態轉變與發育路徑，適合分化或疾病進程研究。
Infer cell state transitions and developmental paths for differentiation or disease progression studies.