Workflow panel
The bottom-right panel maps papers onto the steps of a scRNA-seq analysis. It has two views, switched with the toggle above the diagram: This paper and All ingested papers.
The workflow steps
| Step | Covers |
|---|---|
| scRNA-seq | The sequencing technology or protocol (e.g. 10x Chromium, CITE-seq, Smart-seq2). |
| Quantification | Turning reads into a count matrix (e.g. Cell Ranger, Alevin). |
| Expression matrix & preprocessing | QC filtering, doublets, normalization, highly variable genes, differential expression / marker selection. |
| Dimensionality reduction | PCA, UMAP, t-SNE, learned embeddings. |
| Batch correction | Integrating samples or datasets (e.g. Seurat anchors, Harmony, scVI). |
| Clustering | Grouping cells (e.g. SNN graph clustering, Louvain, Leiden). |
| Annotation | Assigning cell type labels; the project's main focus. |
| Trajectory | Ordering cells along differentiation or time. |
| Cell–cell communication | Ligand–receptor interaction analysis. |
"This paper" view
Each step is colored by how the selected paper relates to it:
| Label | Meaning |
|---|---|
| Core contribution (dark) | The paper's main contribution is at this step. |
| Used (light) | The paper performs this step with existing methods. |
| Mentioned (dashed outline) | Discussed, assumed or only shown, not performed. |
| Not addressed (faint) | The paper doesn't cover this step. |
| Unknown (dotted) | The agent couldn't tell from what it read, usually because only the abstract was available. |
Click a step (or Tab to it and press Enter) to see a short description of what the paper does there, plus the methods and tools it names, and every algorithm and tool in the paper with its role: proposed (the paper's own), used, or compared (a baseline).
Tool and method names with a highlight on hover are clickable: they open that tool's card in the "All ingested papers" view.
"All ingested papers" view
The same diagram, now summarizing every ingested paper:
- Shading: the darker a step, the larger the share of papers that perform it (core or used).
- Box label: e.g. 9/13 papers · Seurat means 9 of 13 papers perform the step and Seurat is its most-used tool. Papers marked unknown for a step are left out of that step's total.
Click a step to see ranked bar charts of its tools and methods (top 12 each). Hover a bar for a breakdown by role; click it to open the tool card. A * after a name means it isn't in the vocabulary yet (vocabulary).
How usage is counted
Choose the measure in the drop-down:
| Measure | Counts | Answers |
|---|---|---|
| Papers using it | Papers that use or propose the tool at this step | What do people actually run? |
| Used as a baseline | Papers that compare against it | What do new methods benchmark against? |
| Proposed here | Papers where it's the paper's own method | Which new methods target this step? |
| Citation-weighted | Sum of the citation scores (0–1) of papers using it | What do the most-cited papers use? |
- Each paper counts once per tool per step, however often the tool is mentioned.
- Mentioned-only uses aren't counted under any measure, so a tool someone only names doesn't get credit as if they ran it.
- Name variants count as one tool: Seurat, Seurat v4 and Seurat v5 are all Seurat. Versions are shown on the tool card.
- Preferred tier only restricts every count (and the coverage numbers) to papers in the preferred tier.
The counts describe the papers ingested so far, which lean toward annotation-method papers. Treat them as a demonstration until many more papers are ingested; see limitations.
Tool cards
A tool card summarizes one tool or method across all ingested papers:
- Its kind (toolkit, database, LLM…), language, the versions seen and a link.
- By workflow step: how many papers use it at each step. Click a step to jump to its chart.
- Papers: every paper that uses it and in what role. Click one to open it.
- Show only these papers in the list: filters the paper list to them. Clear it with the × on the chip above the list.
Use ← Back to return to the step's charts.