Comprehensive Reference Guide for the OPS Explorer

Use the Explorer

The OPS Explorer is a full-screen, multi-panel workspace for exploring OPS and CROP-seq datasets together. You can view embeddings, fluorescence image galleries, and volcano plots side by side. Everything is linked, so selecting a gene knockout highlights it across every open panel simultaneously.

A collection may contain both OPS datasets (morphological imaging) and CROP-seq datasets (single-cell transcriptomics). The viewer can display panels from both modalities at once, and each panel type behaves slightly differently depending on the underlying data.

Without downloading anything you can:

• Browse collections and per-experiment metadata
• Open the volcano plot for any visualization to see which perturbations move which features
• Open the UMAP/embedding to see how perturbations cluster
• View representative single-cell image crops for any perturbation
• Inspect the schema/feature definitions before you write code against them

Contents

• Opening the Viewer
• Collection Info View
• Default Layout
• The Left Panel
  • Gene Knockouts
  • Gene
  • Cluster
  • Layout Tab
  • Adding a Visualization Panel
  • Editing or Removing a Panel
• Visualization Panes
  • Embedding Panel
  • Images Visualization Pane
  • OPS Feature Volcano Plot
  • CROP-seq Gene Expression Volcano Plot
  • Annotations Tab
  • Gene Expression Tab
• Interacting with the Viewer
  • Hovering a Dot
  • Cross-Panel Sync
• Quick Reference

Opening the Viewer

Click Explore on one of the collections surfaced on the OPS Explorer About page to open

Collection options: three buttons in the top-right of the left panel next to the collection name:

• Collection details ⓘ - replaces left panel content with a collection summary
• Download collection ⤓ - provides code snippet to download the collection via CLI
• Resources 📖 - opens a dropdown with 2 actions:

Collection Info View

Click ⓘ Collection details to replace the Gene Knockouts and tab content with a collection-level summary.

The summary includes:

• ← Back button - returns to the Gene Knockouts view
• Collection title
• Publication - linked citation
• Contact - linked researcher name
• Description of experimental context
• Download button

The viewer always shows one collection at a time. A collection may contain multiple OPS datasets, multiple CROP-seq datasets, or both. Each open panel shows data from one dataset within that collection.

Default Layout

What you see on first open depends on which dataset type you launched from:

Launched from	Default panels
OPS dataset	Embedding (left) + Images (right) - side by side
CROP-seq dataset	Embedding only - full width

The OPS default gives you a ready-made view: as soon as you select a gene knockout, the Images panel populates automatically.

Each panel shows a floating label in its top-left corner with the dataset name and, for embeddings, the embedding name (e.g. UMAP_marker_x, marker_y). The Images panel shows only the dataset name.

You can change this layout at any time using the Layout Tab.

The Left Panel

The left panel is the control center for the entire viewer. It is divided into two regions:

• Gene Knockouts (top half) - search and select knockouts to highlight across all panels
• Three tabs (bottom half) - Layout, Annotations, Gene Expression

Gene Knockouts

The Gene Knockouts (KO) section is the primary control. Selecting a knockout there highlights it in every open visualization simultaneously. The embedding, the images panel, and the volcano plot all update together.

To add a gene knockout:

1. Click the Search for gene knockout(s) bar and search for your gene knockout of interest.
2. A dropdown lists all knockouts available in the collection. Checked items are already selected.
3. Click a name to add it, or uncheck to remove it.
4. If your search term has no results, No matches found appears.

Once added, each knockout appears as a row with:

• The gene name
• An ⓘ button - opens the Gene Info sidebar
• A 🗑️ button - removes it from the selection

A counter in the top-right shows how many are active.

Hovering a name in the list highlights that knockout's dot across all open panels and fades all other dots into the background.

Gene

After clicking ⓘ next to a gene knockout:

The sidebar shows:

• Gene name and description
• External database links: Gene Pathways ↗ and Protein Complexes ↗
• A Knockout Effects table with physical and transcriptional effect summaries for this gene in the current dataset

Click − to minimize the sidebar or × to close it.

Cluster

Clicking an ⓘ button next to a cluster in the Annotations Tab opens a cluster info sidebar.

The sidebar shows a breadcrumb identifying the dataset, embedding, and cluster, followed by a table listing all gene knockouts assigned to that cluster.

Layout Tab

The Layout tab controls which panels are open and how they are arranged. It is the default tab when you first open the viewer.

Visualization list - each row represents one open panel and shows:

• A thumbnail indicating the panel's position in the layout
• A chip label: Embedding, Images, or Volcano Plot
• The dataset name as a subtitle
• Edit (✏️) and Delete (🗑️) buttons

+ Add Visualization - opens the Add Visualization sidebar. The viewer supports a minimum of 1 and a maximum of 4 panels open at once. See Adding a Visualization Panel.

Adding a Visualization Panel

Click + Add Visualization in the Layout Tab. A sidebar slides open on the left.

Step 1 - Choose a visualization type:

Step 2 - Select a dataset from the dropdown.

Step 3 - Select an embedding (Embedding and Volcano Plot types only). Multiple embeddings may be available per dataset.

Note: Image panes are available for OPS datasets only. No embedding selection is needed.

Once all required fields are complete, click Add Visualization. The sidebar closes and the new pane appears in the layout.

Editing or Removing a Panel

In the Layout Tab, each row has the following action buttons:

• Edit (✏️ icon) - opens the same Add Visualization sidebar pre-filled with the current settings. The confirm button reads Apply Changes.
• Delete (🗑️ icon) - removes the panel immediately.

Visualization Panes

The right side of the viewer shows one to four panels. This section covers what each panel type displays and how to interact with it. For hover and click interactions shared by all panels, see Interacting with the Viewer.

Embedding Panel

The Embedding panel shows a scatter plot where each dot is one gene knockout, positioned according to the dataset authors' chosen embedding. For OPS datasets, positions typically come from morphological imaging features (typically computed with CellProfiler or a deep learning model; e.g. nucleus size). For CROP-seq datasets, positions come from transcriptomic profiles.

When gene knockouts are selected from the Gene Knockouts section, they appear as labeled, highlighted dots on the plot. See Hovering a Dot and Clicking a Dot for interaction details.

Images Visualization Pane

The Images pane shows representative fluorescence image crops for each selected gene knockout, organized as a grid of rows (one per knockout) and columns (one per imaging marker; e.g., DAPI, ConA, Phalloidin).

When gene knockouts are selected:

Grid structure:

• Column headers - each column is one fluorescent marker, a stain highlighting a specific cellular structure (e.g. DAPI for DNA, ConA for cell membranes, phalloidin for actin)

• CONTROL row - always pinned at the top: it stays fixed as you scroll down, so you always have control images as a visual reference

• Gene KO rows - one row per selected knockout, with approximately 3 representative image crops per marker
• Each row can be collapsed or expanded with the arrow on its header

Image display modes

• Markers: N ▾ dropdown (top-right) - select how many markers to show as columns. Up to 20 markers can be displayed at once. Scroll horizontally to see all columns.
• Hovering an individual image shows information confirming the gene KO name and marker
• Clicking an image opens it in the idetik viewer in a new browser tab for full-resolution viewing

The images pane shows image crops of selected gene knockouts:

OPS Feature Volcano Plot

The OPS Feature Volcano Plot shows the morphological effect of every gene knockout in the screen for a chosen cellular feature. This is useful for identifying which perturbations have the strongest or most statistically significant effects.

What the dots represent:

• Each dot = one gene knockout (same as the Embedding Panel)
• Selected gene knockouts are labeled on the plot
• Dots can be colored by Annotations

Axes:

• X-axis: Effect Size - magnitude of morphological change vs. non-targeting control for the selected feature
• Y-axis: −log₁₀FDR - false discovery rate corrected p-value as a measure of statistical significance

Feature selector

• top-right dropdown - choose which morphological feature drives the x-axis (e.g. Nucleus Size, Nucleus Area, Mitochondria Intensity). The selected feature name also appears on the x-axis label.

For hover and click interactions, see Interacting with the Viewer.

CROP-seq Gene Expression Volcano Plot

The CROP-seq Gene Expression Volcano Plot shows which transcripts are differentially expressed for a chosen gene condition.

Note: Each dot here is a measured transcript (gene), not a gene knockout. There are approximately 20,000 dots, one per transcript in the CROP-seq dataset.

Comparison with the OPS Feature Volcano:

	OPS Feature Volcano	CROP-seq Gene Expression Volcano
Each dot is	A gene knockout	A measured transcript
Dot count	~1,000–5,000	~20,000
Selected KOs labeled?	Yes	No - dots are different entities
Colored by annotation?	Yes	No
Selector changes	Morphological feature	Measured gene
X-axis	Effect Size	Log₂ Fold Change
Y-axis	−log₁₀FDR	−log₁₀(p-value)

Gene selector (top-right dropdown): choose which measured gene to use to display a volcano plot. The plot then shows how the expression of all ~20,000 transcripts changes when that gene is knocked out.

Annotations Tab

The Annotations tab lets you color all eligible panels by a biological annotation, making it easy to see how gene knockouts group by cluster or control status. All annotations are supplied by the dataset authors.

Shared Categories apply across all datasets in the collection:

• control type - control and non-control, each with a cell count

Per-dataset sections appear below (e.g. OPS pseudobulked IF7_432 / UMAP_marker_x, marker_y):

• clusters - cluster 1 through cluster 5+, each with a count and an ⓘ button that opens Cluster

To apply a color: click the filled droplet (💧) button next to any category or clusters label. All eligible panels update immediately.

Which panels get colored:

Panel type	Gets annotation color?
OPS Embedding	Yes
CROP-seq Embedding	Yes
OPS Feature Volcano Plot	Yes
CROP-seq Gene Expression Volcano Plot	No - dots are measured genes, not knockouts

With annotation color and hover active:

Hovering a dot highlights its cluster and fades all other clusters across all linked panels simultaneously.

With cluster labels on:

Gene Expression Tab

The Gene Expression tab lets you color panels by the expression level of a specific measured gene. This coloring applies only to CROP-seq panels.

To add a gene expression overlay:

1. Click Add gene(s) and type a gene symbol.
2. The gene appears as a row with a miniature expression histogram.
3. Click the 💧 droplet icon on that row to activate it as the color overlay.

CROP-seq embeddings are colored with a continuous (gradient) color scale based on expression level.

OPS panels show an amber ⚠ Limited badge. Hovering the badge shows:

Switching back to the Annotations Tab restores annotation-based coloring across all eligible panels. If more than 20 gene knockouts are selected there will be a pop-up notification Gene knockout limit reached.

Interacting with the Viewer

This section covers hover, click, and cross-panel behaviors that apply across all panel types.

Hovering a Dot

Hovering any dot in an embedding or volcano panel shows a tooltip. The fields shown depend on the panel type:

OPS or CROP-seq Embedding, OPS Feature Volcano:

Field	Description
Gene KO	Name of the gene knockout
Cluster	Cluster assignment
# Cells Observed	Number of cells that received this perturbation
Images	Number of representative images available (embeddings only)

When annotation color is active: hovering also fades all dots not in the same cluster, isolating the hovered cluster across all eligible linked panels at the same time.

CROP-seq Gene Expression Volcano: tooltip shows the transcript name, log₂FC (fold change), and −log₁₀FDR (false discovery rate = corrected p-value).

Hovering a name in the gene list: highlights that knockout's dot across all open panels and fades all other dots into the background.

Cross-Panel Sync

When you hover a dot in any embedding or volcano panel, the same gene knockout is simultaneously highlighted in all other open panels. This lets you compare how the same perturbation appears across the OPS embedding, CROP-seq embedding, OPS volcano, and images without navigating away.

Orientation switcher - Icons at the top change the panel arrangement depending on how many visualization panes are open:

Number of open panels	Orientation options
1	Fixed - no choice
2	Side by side or stacked
3	4 arrangement options
4	Fixed 2×2 grid - no choice

Quick Reference

Action	What to do
Find a gene knockout in the embedding	Type the name in the Gene Knockouts search bar
Select feature for OPS volcano	Feature selector dropdown in top right of volcano panel
See microscopy images for a knockout	With Images pane open in the viewer, select a gene knockout and the Images pane will populate automatically
Open an image at full resolution	Click any image in the Images Pane - opens in the idetik viewer
Identify knockouts with the largest morphological effect	Look at the top edges of the OPS Feature Volcano Plot
Change which feature the volcano shows	Use the Feature selector dropdown in the volcano panel header
See which transcripts change for a knockout condition	Open a CROP-seq Gene Expression Volcano; use the Gene selector
Color all panels by cluster	Annotations Tab → click droplet icon (💧) next to clusters
See which knockouts share a cluster	Hover any dot while annotation color is active - other clusters fade
Locate a knockout across all open panels	Hover its dot in one panel - all other panels highlight it simultaneously
See all knockouts in a cluster	Click ⓘ next to an annotation category in the Annotations tab → Cluster Info sidebar
Look up external gene info	Click ⓘ next to a name in the Gene Knockouts list → Gene Info sidebar
Add a new panel	Layout Tab → + Add Visualization
Swap two panels' positions	Click the other rectangle in a panel's thumbnail in the Layout Tab
Download the collection	Click ⤓ next to collection name → command to Download data via our CLI
View collection publication and metadata	Click ⓘ to the right of collection name at top of Layout panel