User manual

Purpose and scope

RCC Explorer is a research workspace for nCounter RCC data analysis. It reads raw Reporter Code Count files, reviews run and lane quality, compares expression patterns, calculates derived normalization layers and prepares documented exports for further work.

nCounter is a targeted digital expression platform. It is included on Spatial Biology because it often sits beside spatial workflows in a research facility, not because nCounter itself is a spatial assay. RCC Explorer does not claim diagnostic use, regulatory validation, manufacturer endorsement or equivalence to nSolver.

Privacy and local processing

Your account is online, but your scientific workspace is local. DigitalOcean serves public pages, authenticated route handlers and the empty Explorer shell, and receives ordinary hosting request metadata such as IP address and request headers. Firebase manages identity, verification, session-token validation, email-domain policy, profiles and minimal account-access events. RCC/ZIP contents, filenames, local project details, labels, counts, QC results, plots and exports stay in the browser or local filesystem and are sent to neither provider.

The browser receives files through the folder permission you grant. It reads them into memory and does not modify the source folder. Exports are ordinary browser downloads. Do not type patient or sample information into account fields, because account fields are online.

Local does not mean backed up. Clearing browser storage can remove the local project catalogue, and closing a tab removes the active analysis state. Keep the source files and important exports in your normal managed storage.

Requirements

• Current desktop Google Chrome or Microsoft Edge. Folder access depends on the File System Access API and is not offered on mobile browsers.
• An internet connection for sign-in, session checks and loading the empty explorer shell.
• A verified organisational or institutional email account. Public and disposable email providers are not accepted unless the operator creates an explicit exception.
• A folder containing direct-child .RCC files, .ZIP archives, or both. Subfolders on disk are not scanned.
• No more than 1,000 supported files and 100 MB in one folder selection. Expanded ZIP content is checked against the same safety limits.

The public manual works on phones and tablets. The analysis workspace is designed for a desktop screen, keyboard and pointer.

Account access

1. Open Create account and enter your name, institution and organisational email address.
2. Use a password of 12 to 128 characters and accept the current terms and privacy notice.
3. Open the Firebase verification email. The verification link returns to the Spatial Biology account-action page.
4. Go to Sign in. A verified sign-in creates a secure, HTTP-only browser session and removes the temporary Firebase browser credential.

If the verification email does not arrive, check spam first. The resend screen asks for the account password because Firebase must authenticate the request. Password-reset requests always show the same neutral response, whether or not an account exists.

When a session expires while the explorer is open, the analysis remains in that tab but its controls lock. Sign in in a new tab, return to the explorer and wait for the session check. Keep the analysis tab open during recovery.

Create and edit a project

1. Open Your projects and choose Create project.
2. Enter a project name and optional description. These fields are stored only in IndexedDB in this browser, scoped to the signed-in account.
3. Choose Choose project folder, select the folder holding the RCC or ZIP files and grant read access.
4. To rename a project later, choose Edit details, change the name or description, then choose Save changes. The folder connection and saved view preferences remain attached.

Removing a project deletes only its browser catalogue entry. It does not delete or alter the folder. The same project catalogue does not automatically appear in another browser or on another computer.

Load RCC and ZIP data

Opening a project reads supported files directly inside its connected folder. Folder scanning is nonrecursive. A ZIP may contain RCC files in folders, and a combined ZIP may contain cartridge ZIPs one level deep.

Supported archive behaviour

• Stored and deflated ZIP entries are supported with size and integrity checks.
• One level of nested ZIP is supported. A ZIP nested two levels deep is not opened.
• Encrypted, multipart and ZIP64 archives are rejected with an explicit message.
• RCC text is decoded as UTF-8, with a Windows-1252 fallback for older files.

Replace or add

Replace earlier uploads validates the new selection before replacing the current in-memory dataset. Invalid input leaves the earlier workspace in place. Add to the current datasets merges records with a compatible CodeSet and probe set. Probe rows are aligned by code class, name and accession. Incompatible CodeSets become separate datasets.

Repeated filenames and sample IDs remain separate records with unique internal keys. They are not interpreted as duplicates automatically.

Source provenance

The source rail lists cartridges in selection order. Choose Clear, then select cartridges in the order required for plots, tables and count exports. Deselecting a cartridge removes it from the active order; selecting it again places it last.

Provenance checks surface unusual relationships such as a CartridgeID shared by several archives, a lane range that disagrees with its CartridgeID, repeated sample IDs, repeated filenames or a lane containing only zero counts. These are review notes, not automatic corrections.

Source labels and raw counts are never changed. If a known laboratory labelling error exists, keep the imported value visible and document your decision separately rather than rewriting the source inside the explorer.

Overview and panel detection

The Overview identifies the dataset, number of archives, lanes and probes, then summarizes the panel and instrument fields found in the RCC records. Detection uses the explicit AssayType when available and otherwise examines code classes and probe names.

• Panel types may be reported as gene expression, miRNA, CNV, protein, PlexSet or unknown.
• CodeSet origin is inferred from catalogue and custom naming patterns and is shown as evidence, not as a guarantee.
• The positive-control ladder and negative-control count are checked for consistency.
• Instrument family is shown as SPRINT, MAX/FLEX or unknown using recorded scanner and field-of-view evidence.
• FOV settings are reported as 280 standard, 555 maximum or mixed where the lanes disagree.

The inferred instrument selects the initial binding-density preset, and a detected miRNA panel selects a top-100 normalization default. Both remain editable. Differences between lanes are shown rather than reconciled silently.

RCC quality control

The RCC quality control view implements the core assay checks used in nCounter review. Thresholds are editable, and a change does not alter the RCC record.

Imaging

Fields of view successfully counted. The published default shown by the tool is at least 75%.

Binding density

Reporter density in the scanned field. Defaults are 0.1 to 2.25 for MAX/FLEX and 0.1 to 1.8 for SPRINT.

Positive-control linearity

Log2 count against log2 concentration for POS_A through POS_E. POS_F is excluded from the regression. The default R-squared threshold is at least 0.95.

Detection limit

POS_E must be above the lane's negative-control mean plus two sample standard deviations under the default setting.

The default cohort leaves out assay-flagged lanes and then checks positive and housekeeping normalization factors. Every default exclusion is listed with a reason. Exclude flagged lanes from cohort is a convenience action, not a deletion. You can include or exclude lanes explicitly.

A QC pass does not prove biological validity, correct sample identity or a sound study design. A QC flag does not prove a sample should be discarded.

Background and detection

Background handling creates a derived layer for endogenous and housekeeping probes. Raw counts remain separate and visible.

• None keeps counts unchanged before positive and content scaling.
• Threshold raises a value below the selected background statistic to that statistic.
• Subtract subtracts the selected background and floors the derived value at one.

Available negative-control summaries include mean, mean plus a chosen number of standard deviations, median, geometric mean and maximum. The default detection limit is calculated per lane as negative mean plus two sample standard deviations.

The detection table reports, for every endogenous and housekeeping gene, how many visible lanes exceed their own lane-specific background. An optional minimum-detection percentage can hide genes from other views and their view-specific exports. It does not remove probes from the raw dataset.

Normalization

The nCounter normalization workflow applies positive-control scaling first, then optional background handling, then a selected content-normalization method. The cohort and reference genes are explicit inputs.

Available content methods

• Reference-gene geometric mean or reference-gene sum.
• Global geometric mean or total endogenous count.
• Top-N geometric mean and low-CV geometric mean.
• Median of ratios, based on the DESeq2 size-factor approach.
• Trimmed mean of M-values, based on edgeR TMM.
• Quantile normalization.
• RUVg using selected reference genes as negative-control genes.
• Positive-control scaling only.

The housekeeper review checks whether candidates are commonly detected above background, have reasonable variation after positive scaling and agree with other candidates using geNorm M. These criteria produce advice. They do not establish that a gene is biologically stable in your study.

Compute comparison reports median CV, relative-log-expression departure, PC1 association with cartridge and repeat-run agreement when applicable. A lower comparison metric does not automatically identify the correct normalization. A method can reduce technical variation while also removing real biology. Use study design, controls and diagnostic plots together.

Samples and groups

An optional CSV sample sheet adds display labels and grouping variables without rewriting RCC attributes. Download the template first if you want the exact current lane identifiers.

filename,sample,group,tissue
synthetic_lane_01.RCC,Example 01,Control,Liver
synthetic_lane_02.RCC,Example 02,Treatment,Liver

The first line may use the minimal header filename,sample,group. A row is matched by RCC filename first. You may instead use an archive column together with a lane column. This is useful when the same RCC filename appears within different cartridge archives.

Rows that do not match remain listed as unmatched. Lanes without a matching row remain visible and unlabelled. Duplicate column names and sheets without a filename or archive-plus-lane identifier are rejected. Group and other extra columns become annotations for colouring plots and for the exported sample sheet.

Plots and visual views

Samples

Search and inspect lane metadata, RCC sample IDs, archives, lane numbers and sample-sheet annotations. Repeated labels stay visible.

Gene view

Select a probe and compare its raw or derived values across the visible lanes. Use the source order and annotations to identify cartridge or group structure.

Heatmap

Choose raw or normalized values, log2, linear or per-gene z-score colour treatment, and gene ordering by RCC order, mean, variance or name. The minimum-count filter keeps a gene when it reaches the threshold in at least one selected lane. Large matrices are paged at 100 genes by 100 lanes.

Sample structure

PCA and lane-to-lane correlation use log2 normalized values and can be coloured by archive or sample-sheet group. PCA is descriptive. Separation can reflect biology, cartridge effects, sample composition or another source of variation.

Plots can be exported as SVG for editable vector graphics or PNG for convenient viewing. Always retain the settings and sample sheet that explain what is shown.

Exports and downstream analysis

Export filenames include the dataset label and date. A raw-count export follows the filters visible in the current view, so check source order, probe-class filter, gene search and detection filter before downloading.

• Raw counts: original RCC integers in the selected table shape.
• QC report: lane metrics, flags and the default-cohort reason.
• Derived counts: background and normalization output with lane flags on every row.
• Detection table: per-gene detection counts and fractions.
• Analysis settings JSON: thresholds, cohort, reference genes, background and normalization choices.
• Sample-sheet template: lane identifiers ready for annotation.
• SVG or PNG plots: the current visual view.
• Analysis bundle ZIP: raw wide matrix, log2 normalized wide matrix, sample sheet, settings and a README with an example limma workflow.

Differential expression belongs in a statistical workflow outside RCC Explorer. The bundle prepares documented inputs for R. Review the design matrix, contrasts, repeated measures, confounders and multiple-testing approach with an appropriate analyst.

Reopen and reconnect

The project catalogue saves the project name, description, folder handle, last-opened time and selected display preferences locally. Reopening a project rereads the source folder, so a changed source file is reflected.

Browser permission may return to a prompt after a restart. Choose Open project and grant read access. If the folder moved, the external drive is disconnected or the saved handle is no longer usable, choose Reconnect folder and select the intended folder again.

Cartridge order, count threshold, gene search, probe class and display scales are saved locally. The active dataset, QC and normalization choices, sample sheet and in-memory results are session-only and must be reapplied after reopening. Export settings when a decision needs a durable record.

Troubleshooting

No supported files found

Select the folder that directly contains RCC or ZIP files. Disk subfolders are not scanned.

Selection exceeds a limit

Reduce the selection to at most 1,000 supported files and 100 MB, including expanded ZIP content.

ZIP rejected

Check whether it is encrypted, multipart, ZIP64, damaged or nested more than one ZIP level deep. Extract it safely and select the supported cartridge files.

Different CodeSets appear

The explorer separates incompatible probe sets. Select the required dataset from the dataset control instead of merging unlike panels.

Folder permission needed

Open the project again and grant read access. If that fails, use Reconnect folder.

Explorer locked

Restore the internet connection or sign in in a new tab. Keep the analysis tab open so the current in-memory work is retained.

Unexpected label or repeated ID

Inspect the provenance note and laboratory record. The explorer deliberately keeps source values unchanged.

Export seems incomplete

Check the visible archives, probe class, search, detection filter and current view. Exports respect their visible scope.

Methods and references

Core assay QC and default normalization logic follow the equations described in NanoString's nCounter gene-expression data-analysis guidance, implemented independently. The tool also labels methods derived from established literature, including geNorm, DESeq2 median-of-ratios, edgeR TMM, quantile normalization, RUVSeq and related nCounter workflows.

• Vandesompele et al. (2002), geNorm reference-gene stability.
• Anders and Huber (2010), median-of-ratios size factors.
• Love et al. (2014), DESeq2.
• Robinson and Oshlack (2010), TMM normalization.
• Bolstad et al. (2003), quantile normalization.
• Risso et al. (2014), RUVSeq.

Method names and citations indicate the computational basis. They do not establish parity with a vendor implementation. Numerical parity with nSolver is not claimed unless a specific export is supplied to the optional parity test.

Glossary

Archive

A selected ZIP source, usually representing a cartridge or a packaged set of RCC files.

CodeSet

The probe definition associated with a lane, including code class, target name and accession.

Cohort

The lanes used to estimate positive and content-normalization references.

Derived count

A value calculated from a raw count after the recorded background and scaling steps.

FOV

Field of view. The number and success of scanned fields contributes to imaging QC.

Lane

One RCC record or well within an nCounter cartridge run.

LOD

Limit of detection, calculated here from the lane's negative controls under the selected rule.

Project

A browser-local name, description and folder connection associated with one signed-in account on one browser.

Raw count

The integer imported from the RCC Code_Summary table before any derived processing.

Reference gene

A selected housekeeping or control gene used to estimate content-normalization factors.