Output files

The output files Cells2Stats produces

The following table lists the files and folders that Cells2Stats outputs. Parquet files are column-based files that efficiently store data. For more information, see the Apache Parquet Documentation.

In the table, {root} is the output directory, which defaults to RUN_DIRECTORY/Cytoprofiling/{timestamp}. {well} is the well folder name, such as WellA1, and {RunName} is the run name from RunParameters.json.

File	Directory and File Name	Description	Quantity
Spatial Data Object	{root}/SpatialData/{RunName}.zarr.zip	Zarr-based structured object containing all modalities from a run for downstream analysis and visualization. See Spatial Data Object.	One per run
CytoCanvas™ Visualization File	{root}/SpatialData/cyto.viz	Manifest that the Element Biosciences CytoCanvas tool opens to load a run. Lists the visualization files and targets.	One per run
CytoCanvas Image Index	{root}/SpatialData/{RunName}.zarr-index.bin	Index over the images in the Spatial Data Object, which lets CytoCanvas read images without extracting the zip	One per run
CytoCanvas Points	{root}/SpatialData/points-zarr.zip	Polony positions and target assignments for display in CytoCanvas	One per run
CytoCanvas Points Index	{root}/SpatialData/{RunName}.points-zarr-index.bin	Index over points-zarr.zip	One per run
CytoCanvas Cell Shapes	{root}/SpatialData/cell-shapes-zarr.zip	Cell boundary shapes for display in CytoCanvas	One per run
CytoCanvas Cell Shapes Index	{root}/SpatialData/{RunName}.cell-shapes-zarr-index.bin	Index over cell-shapes-zarr.zip	One per run
CytoCanvas Nuclear Shapes	{root}/SpatialData/nuclear-shapes-zarr.zip	Nuclear boundary shapes for display in CytoCanvas	One per run
CytoCanvas Nuclear Shapes Index	{root}/SpatialData/{RunName}.nuclear-shapes-zarr-index.bin	Index over nuclear-shapes-zarr.zip	One per run
Run Stats	{root}/RunStats.json	JSON file that records run, well, batch, and target statistics for every assay in the run. See RunStats.json.	One per run
Raw Cell Statistics, Parquet	{root}/RawCellStats.parquet	Parquet file that reports per-cell position, size, morphology, target counts, and intensity values. See RawCellStats.parquet.	One per run
Average Normalized Well Statistics	{root}/AverageNormWellStats.csv	Filtered and average metrics for each well in the run	One per run
Versions, JSON	{root}/Versions.json	File that reports the version number for CSV output files and bundled software programs. See Versions.	One per run
QC Report, HTML	{root}/qc_report.html	Interactive HTML quality control (QC) report for the run. See QC report.	One per run
Run Manifest, CSV	{root}/RunManifest.csv	Run manifest for the Cells2Stats execution	One per run
Run Manifest, JSON	{root}/RunManifest.json	JSON file that is reserved for Element processes	One per run
Run Parameters	{root}/RunParameters.json	JSON file that records information about the run configuration	One per run
Panel	{root}/Panel.json	JSON file that records information about the targets for the run	One per run
Cell Segmentation Mask	{root}/CellSegmentation/{well}/{tile}_Cell.tif or {tile}_Cell.zarr	Cell segmentation masks for a well, where the Cell ID is the value for a pixel in a cell. The format follows the run, as described in Segmentation mask formats.	One per tile per well
Nuclear Segmentation Mask	{root}/CellSegmentation/{well}/{tile}_Nuclear.tif or {tile}_Nuclear.zarr	Nuclear segmentation masks for a well, where 1 is the value for a pixel in a nucleus. The format follows the run, as described in Segmentation mask formats.	One per tile per well
Corrected Projection Image	{root}/Projection/{well}/{batch}_{tile}_{target}.tif	Corrected projection images of imaging targets. Present only when --process-imaging-targets is used.	One per target per tile
Log Files	{root}/Logs/	Cells2Stats.log, manifest error files such as RunManifestErrors.json, and logs from bundled tools	One folder per run
Barcodes	{root}/Wells/{well}/{batch}/{tile}_barcodes.parquet	(Barcoding) Parquet files that provide barcoding information for each polony in a tile. See Barcoding output files.	One per tile per batch per well
Guide RNA Reads	{root}/Wells/{well}/{batch}/{tile}_reads.parquet	(Guide RNA) Parquet files that provide alignment and cell assignment information for each polony in a tile. See Guide RNA output files.	One per tile per batch per well
Target Counts, JSON	{root}/SpecializedTargetedCounts.json	(Guide RNA) JSON file that records read counts per spacer target across the run	One per run
Target Cell Assignment Manifest, CSV	{root}/TargetCellAssignmentManifest.csv	(Guide RNA) Copy of the target cell assignment manifest used in this run	One per run
Target Cell Assignment Manifest, JSON	{root}/TargetCellAssignmentManifest.json	(Guide RNA) JSON file that is reserved for Element processes	One per run
Guide RNA Alignments, BAM	{root}/Wells/{well}/{batch}/{well}_{targetSite}_alignments.bam	(Guide RNA) Concatenated alignments for a target site. Present only when --generate-specialized-targeted-bams is used.	One per target site per batch per well
3′ Transcriptome Reads	{root}/Wells/{well}/{batch}/{tile}_rawreads.parquet	(3′ Transcriptome) Parquet files that provide position and cell assignment information for each polony in a tile. See 3′ Transcriptome output files.	One per tile per batch per well

Key output files

Most downstream analyses start from three files:

• Spatial Data Object: All modalities from the run in a single structured object.
• RunStats.json: Run-level and well-level quality metrics for each assay.
• RawCellStats.parquet: One row per cell, with position, size, and per-assay measurements.

Spatial Data Object

SpatialData/{RunName}.zarr.zip is a Zarr-based structured object that contains all modalities from the run, including images, segmentation labels, cell and nuclear shapes, polony points, and a per-cell table. It is the primary output for downstream analysis and visualization. The CytoCanvas visualization files are written next to it in SpatialData/.

For the structure and contents of the Spatial Data Object, see Spatial Data Object.

RunStats.json

RunStats.json reports statistics at the run, well, batch, and target levels. Each assay writes its statistics to a dedicated top-level section:

Top-level section	Assay
CytoStats	General, Cell Paint, and Barcoding
ImagingTargetsStats	Cell Paint and Protein
DemuxStats	Barcoding
SpecializedTargetedStats	Guide RNA
ThreePrimeUntargetedStats	3′ Transcriptome

A section for an assay that is not in the run is present but contains no batches. A metric that cannot be calculated is reported as -999.

General

The following fields describe the run and cell population, independent of assay:

Field	Location	Description	Data Type
AnalysisID	Top level	Unique ID for the analysis	String
AnalysisVersion	Top level	Version of the analysis software	String
FileVersion	Top level	Version of the file format	String
FlowCellID	Top level	ID of the flow cell that is used	String
RunID	Top level	Unique ID for the run	String
RunName	Top level	Name of the run	String
CellCount	CytoStats, CytoStats.Wells	Number of segmented cells in the run or well	Integer
Wells	CytoStats	List of well-specific statistics	Array of Objects
WellLocation	CytoStats.Wells	Location of the well, for example A1	String
MedianCellDiameter	CytoStats.Wells	Median diameter of a cell after cell segmentation, in micrometers	Float

Cell Paint

Cell Paint contributes segmentation quality metrics to CytoStats and intensity metrics to ImagingTargetsStats.

The following segmentation metrics are reported at the run level of CytoStats and for each entry in CytoStats.Wells:

Field	Description	Data Type
PercentConfluency	Percentage of the analyzed well area covered by segmented cells	Float
PercentNucleatedCells	Percentage of cells that contain a segmented nucleus	Float
PercentVisualConfluency	Percentage of the analyzed well area that is foreground in the primary cell boundary image, by Otsu thresholding	Float
CellBoundaryPPV	Positive predictive value of the cell masks: the share of bright pixels in the cell boundary target image that fall inside a cell	Float
CellBoundaryBalancedAccuracy	Balanced accuracy of the cell masks at separating cell boundary target signal from background	Float
SupplementaryCellBoundaryPPV	Positive predictive value of the cell masks for the supplementary cell boundary target	Float
SupplementaryCellBoundaryBalancedAccuracy	Balanced accuracy of the cell masks for the supplementary cell boundary target	Float
NuclearPPV	Positive predictive value of the nuclear masks for the nuclear target	Float
NuclearBalancedAccuracy	Balanced accuracy of the nuclear masks for the nuclear target	Float

The segmentation targets are set in the panel. A metric for a segmentation target that the panel does not define is reported as -999.

ImagingTargetsStats reports intensity metrics for each Cell Paint target with the following structure:

ImagingTargetsStats
└── Batches[]
    ├── BatchName
    └── Wells[]
        ├── WellLocation
        └── ImagingTargets[]
            ├── Target
            ├── CellSegmentationTarget
            └── Metrics[]
                ├── MetricName
                └── MetricValue

MetricName	Description
MeanIntensity	Mean intensity of the target inside cells
NuclearMeanIntensity	Mean intensity of the target inside nuclei
ExtracellularMeanIntensity	Mean intensity of the target outside cells
MeanIntensityAboveThreshold	Mean intensity of the pixels above the intensity threshold
NuclearRatio	Share of the above-threshold area of cells that lies within the nucleus

Per-batch background targets, Bkg_RED and Bkg_GRN, and per-target background targets, Bkg_{target}, appear as imaging targets when they are quantified. For details, see Intensity quantification columns.

Protein

ProteinIF imaging targets are reported in ImagingTargetsStats with the same structure and metrics as Cell Paint targets. See Cell Paint.

Protein contributes no other fields to RunStats.json. Barcoded protein targets are reported under Barcoding.

Guide RNA

Guide RNA statistics are reported in SpecializedTargetedStats with the following structure:

SpecializedTargetedStats
└── Batches[]
    ├── BatchName
    ├── PolonyAssignmentStats
    │   ├── PercentAssigned, PercentMismatch, PercentPerfect
    │   └── TargetSitePolonyAssignmentStats[]
    │       ├── TargetSiteName, IsControl
    │       ├── PercentAssigned, PercentMismatch, PercentPerfect, PercentTrimmed
    │       └── UnassignedSequences[]
    │           ├── Sequence
    │           └── Count
    ├── TargetSites[]
    │   ├── TargetSiteName, IsControl
    │   └── CellAssignmentStats
    └── Wells[]
        ├── WellLocation
        ├── PolonyAssignmentStats
        └── TargetSites[]

Field	Description	Data Type
PercentAssigned	Percentage of polonies that are assigned to a target	Float
PercentMismatch	Percentage of assigned polonies with at least one mismatch to the target sequence	Float
PercentPerfect	Percentage of assigned polonies that match the target sequence exactly	Float
PercentTrimmed	Percentage of polonies at the target site that are trimmed	Float
TargetSiteName	Name of the target site as provided in Panel.json	String
IsControl	Whether the target site is a control site. Control sites receive polony assignment only.	Boolean
UnassignedSequences	Most frequent sequences detected at the target site that do not match any manifest entry	Array of Objects
CellAssignmentStats	Cell assignment metrics for the target site. See Target cell assignment metrics.	Object

3′ Transcriptome

3′ Transcriptome statistics are reported in ThreePrimeUntargetedStats.ReadStats with the following structure:

ThreePrimeUntargetedStats
└── ReadStats
    ├── NumPoloniesBeforeTrimming
    └── Wells[]
        ├── WellLocation
        ├── NumPoloniesBeforeTrimming
        └── Batches[]
            ├── BatchName
            ├── NumPoloniesBeforeTrimming
            ├── NumPoloniesPerCellBeforeTrimming
            ├── NumPoloniesPerMM2BeforeTrimming
            └── ExtraCellularRatioBeforeTrimming

Field	Description	Data Type
NumPoloniesBeforeTrimming	Number of pass-filter polonies in the run, well, or batch, counted before trimming	Integer
NumPoloniesPerCellBeforeTrimming	Mean number of pass-filter polonies per cell	Float
NumPoloniesPerMM2BeforeTrimming	Number of pass-filter polonies per mm² of cell area	Float
ExtraCellularRatioBeforeTrimming	Ratio of extracellular to intracellular polony density	Float

Barcoding

Barcoding contributes demultiplexing statistics to DemuxStats and cell-level count statistics to CytoStats.

DemuxStats has the following structure:

DemuxStats
├── NumPolonies, PercentAssignedReads, PercentMismatch
└── Batches[]
    ├── BatchName, NumPolonies, PercentAssignedReads, PercentMismatch
    ├── Targets[]
    │   └── TargetName, ExpectedSequence, NumPolonies, PercentMismatch
    ├── Wells[]
    │   └── WellLocation, NumPolonies, PercentAssignedReads, PercentMismatch
    └── UnassignedSequences[]
        └── Count, Sequence

Field	Description	Data Type
NumPolonies	Number of polonies detected in the run, batch, or well, or assigned to the target	Integer
PercentAssignedReads	Percentage of reads that are assigned to targets	Float
PercentMismatch	Percentage of assigned reads with at least one base pair mismatch	Float
TargetName	Name of the barcoding target	String
ExpectedSequence	Expected barcode sequence for the target	String
Sequence	Barcode sequence of unassigned reads	String
Count	Number of times the unassigned sequence appears	Integer

CytoStats reports the following barcoding fields:

Field	Location	Description	Data Type
AverageAssignedCountsPerMM2	CytoStats	Assigned counts per mm² of cell area for all barcoding targets, averaged across wells	Float
MeanAssignedBarcodingCountPerCell	CytoStats.Wells	Mean number of polonies assigned to a barcoding target per cell	Float
AssignedCountsPerMM2	CytoStats.Wells, CytoStats.Wells.Batches	Assigned counts per mm² of cell area for all barcoding targets	Float
Batches	CytoStats.Wells	List of barcoding batch statistics for the well	Array of Objects
BatchName	CytoStats.Wells.Batches	Name of the batch	String
MeanAssignedCountPerCell	CytoStats.Wells.Batches, target lists	Mean number of assigned polonies per cell	Float
ExtraCellularRatio	CytoStats.Wells.Batches	Ratio of extracellular to intracellular assigned target density	Float
PercentCellularAssigned	CytoStats.Wells.Batches	Percentage of assigned reads that are inside cells	Float
ControlTargets	CytoStats.Wells.Batches	Statistics for quality control targets, with TargetName, ControlType, MeanAssignedCountPerCell, and AssignedCountPerMM2	Array of Objects
AssignedCountPerMM2	Target lists	Assigned counts per mm² of cell area for the target	Float
ManifestLabel	AntibodyTargets	Label of the custom protein in the manifest	String
NuclearFraction	AntibodyTargets	Fraction of the cellular assigned reads of the target that are in the nucleus	Float

RawCellStats.parquet

RawCellStats.parquet contains one row per cell. Assay-specific columns are named {target}{suffix}.{batch} or {column}.{targetSite}.{batch}, so a column name identifies its batch. The Parquet file metadata records FileVersion and AnalysisID.

Note:

Cells2Stats also writes {root}/RawCellStats.csv, which contains the same information in CSV format.

General

The following columns are present for every run:

Column	Description
Cell	Cell ID
Well	Well location, for example A1
WellLabel	Well label from the run manifest
Tile	Tile that contains the cell
X, Y	Cell position in the tile, in pixels
Xum, Yum	Cell position in the tile, in micrometers
WellX, WellY	Cell position in the well, in pixels
WellXum, WellYum	Cell position in the well, in micrometers
Area	Cell area, in pixels
AreaUm	Cell area, in µm²
NuclearArea	Nuclear area, in pixels
NuclearAreaUm	Nuclear area, in µm²

Cell Paint

Cell Paint targets contribute intensity quantification columns and, unless --skip-cellprofiler is used, CellProfiler morphology columns.

Intensity quantification columns

For each imaging target in each batch, RawCellStats.parquet reports intensity quantification columns named {target}{suffix}.{batch}. The suffix identifies the compartment and the measurement:

Suffix	Measurement
None	Mean intensity across the whole cell
_Nuclear	Mean intensity in the nuclear compartment
_Cytoplasmic	Mean intensity in the cytoplasmic compartment
_FractionAboveThreshold	Fraction of whole-cell pixels above the intensity threshold
_FractionAboveThreshold_Nuclear	Fraction of nuclear pixels above the intensity threshold
_FractionAboveThreshold_Cytoplasmic	Fraction of cytoplasmic pixels above the intensity threshold
_MeanIntensityOfFractionAboveThreshold	Mean intensity of the whole-cell pixels above the threshold
_MeanIntensityOfFractionAboveThreshold_Nuclear	Mean intensity of the nuclear pixels above the threshold
_MeanIntensityOfFractionAboveThreshold_Cytoplasmic	Mean intensity of the cytoplasmic pixels above the threshold
_NuclearRatio	Share of the above-threshold area of the cell that lies within the nucleus

_NuclearRatio is empty for a cell with no above-threshold pixels, because the ratio is undefined.

Per-batch background intensity is quantified as the synthetic targets Bkg_RED and Bkg_GRN. To omit them, run analyze with --disable-per-batch-background. Per-target background is reported as Bkg_{target} when RecipeValues.SavePerTargetBkgImages is true in RunParameters.json.

CellProfiler morphology columns

For each imaging target other than background targets, RawCellStats.parquet reports metrics from the following CellProfiler modules:

• MeasureObjectSizeShape
• MeasureGranularity
• MeasureObjectIntensity
• MeasureObjectIntensityDistribution
• MeasureTexture

Certain metrics from these modules are not reported. For example, Zernike metrics and Z-axis metrics are not included.

For more information on these metrics, see the measurement information in the CellProfiler Manuals.

Protein

ProteinIF imaging targets contribute the same intensity quantification and CellProfiler morphology columns as Cell Paint targets, named with the ProteinIF target and batch. See Cell Paint.

Protein contributes no other columns to RawCellStats.parquet. Barcoded protein targets are reported under Barcoding.

Guide RNA

For each target site in each Guide RNA batch, RawCellStats.parquet reports the following columns:

Column	Description
MostAbundantTarget.{targetSite}.{batch}	Target with the most assigned polonies in the cell. NA for control target sites.
MostAbundantTargetCount.{targetSite}.{batch}	Number of polonies in the cell assigned to the most abundant target
TotalCount.{targetSite}.{batch}	Total number of polonies in the cell assigned to any target at the target site
Confidence.{targetSite}.{batch}	Assignment confidence, calculated as MostAbundantTargetCount / (TotalCount + 1). Empty for control target sites.
Gene.{targetSite}.{batch}	Gene of the most abundant target, from the target cell assignment manifest
ControlType.{targetSite}.{batch}	Control type of the most abundant target, from the target cell assignment manifest

For how these values determine pure, mixed, and unassigned cells, see Cell assignment logic.

3′ Transcriptome

3′ Transcriptome contributes no columns to RawCellStats.parquet. Per-polony data is available in the _rawreads.parquet files and the Spatial Data Object.

Barcoding

For each target in each barcoding batch, RawCellStats.parquet reports the following columns:

Column	Description
{target}.{batch}	Number of polonies in the cell assigned to the target
{target}_Nuclear.{batch}	Number of polonies in the nucleus assigned to the target

Assay-specific output files

The following sections describe output files that Cells2Stats writes only when the run includes a particular assay.

Cell Paint

When --process-imaging-targets is used, Cells2Stats writes corrected projection images of the imaging targets to {root}/Projection/{well}/{batch}_{tile}_{target}.tif and uses them for the rest of the analysis. The run directory Projection/ folder is not modified.

Cell Paint has no other assay-specific output files. Cell Paint results are reported in the key output files and the QC report.

Protein

ProteinIF imaging targets are included in the corrected projection images described in Cell Paint. Protein has no other assay-specific output files.

Guide RNA

When target cell assignment is enabled for at least one target site in a run, Cells2Stats populates SpecializedTargetedStats in RunStats.json and generates the following additional output files. For more information on how cell assignment works, see Cell assignment logic.

Cells are classified as pure (single spacer detected), mixed (multiple spacers, one dominant), or unassigned based on a confidence score. For a full explanation of these categories and how cell assignment is calculated, see Pure cells vs. mixed cells.

Key target cell assignment output files

File	What to use it for
SpecializedTargetedCounts.json	See total reads detected per spacer. High dropout (many spacers with zero reads) may indicate a manifest or TargetMask issue. Control targets marked with IsControl receive polony assignment only and do not appear as cell-assigned guides.
TargetCellAssignmentManifest.csv	Confirm which sequences and settings were applied. Compare against your source library file to catch discrepancies.
Panel.json	Contains the TargetMask that was applied per target site. Check here if assignment rates are unexpectedly low.
RunStats.json	Contains all target cell assignment-specific quality metrics. See Target cell assignment metrics.
RawCellStats.parquet	Per-cell assignment and counts at each target site. See Guide RNA columns.
Wells/{well}/{batch}/{tile}_reads.parquet	Per-polony alignment and cell assignment. See Guide RNA reads files.
Wells/{well}/{batch}/{well}_{targetSite}_alignments.bam	Alignments for a target site, for inspection in standard BAM tools. Present only when --generate-specialized-targeted-bams is used.

Guide RNA reads files

Each {tile}_reads.parquet file contains one row per polony, with the following columns:

Column	Description
ReadID	Read identifier. See Read identifiers.
PolonyIndex	Polony index within the tile
X, Y	Polony position in the tile, in pixels
Xum, Yum, Zum	Polony position, in micrometers
WellX, WellY, WellXum, WellYum	Polony position in the well, in pixels and micrometers
TargetName	Target that the polony is assigned to
Cell	ID of the cell that contains the polony
IsNuclear	Whether the polony is inside the nucleus
MapQ	Mapping quality of the alignment
AlignmentLength	Length of the alignment
NM	Edit distance between the read and the target
IsSoftclipped	Whether the alignment is soft-clipped

Target cell assignment metrics in RunStats.json

The following fields are reported in each CellAssignmentStats object in SpecializedTargetedStats, at the batch and well levels. UnassignedSequences is reported in each TargetSitePolonyAssignmentStats object.

Field	Description	What to watch for
PercentAssignedPureCells	Cells assigned to a single dominant spacer (confidence >0.5, meets minimum read count)	Primary metric — aim as high as possible for your MOI
PercentAssignedMixedCells	Cells with multiple spacers detected, but one is dominant (confidence >0.5)	Expected to increase with higher MOI; still valid data
PercentUnassignedMixedCells	Cells with multiple spacers; no spacer exceeds 50% confidence	High values suggest MOI may be too high for reliable single-guide assignment
PercentUnassignedLowCountCells	Spacer was matched but read count is below the minimum threshold	High values suggest low expression or insufficient sequencing depth
PercentTargetDropout	Percentage of spacers in the manifest not assigned to any cell	High dropout may indicate library representation problems
AssignedCountsPerMM2	Assigned counts per mm² of cell area at the target site	Indicator of overall guide detection density
MeanAssignedCountPerCell	Mean number of assigned polonies per cell at the target site	Indicator of per-cell sequencing depth at target site
MedianAbundantTargetCount	Median read count of dominant spacer in assigned cells	Indicator of per-cell sequencing depth at target site
MeanUniqueTargetsPerCell	Mean number of unique spacers detected per cell	Values substantially above 1 suggest high MOI
ExtraCellularRatio	Ratio of extracellular to intracellular assigned polony density	High values suggest signal outside cells or segmentation issues
UnassignedSequences	List of sequences detected at target site but not matching any manifest entry	Primary troubleshooting resource — see Using UnassignedSequences for troubleshooting

Using UnassignedSequences for troubleshooting

The UnassignedSequences field lists actual nucleotide sequences of reads detected at the target locus that did not match any spacer in the manifest. Inspect this field when assignment rates are low.

What you see	Likely cause	Resolution
Sequences that look like your spacers but reversed	Orientation error in manifest	Re-enter spacers in 5' to 3' direction
Spacer-like sequences of a different length	TargetMask or sequence length mismatch	Confirm the Y cycle count in TargetMask equals spacer length
Recognizable spacers from your library	Spacers missing from manifest	Add missing entries to [TARGET INFORMATION]
Unrecognizable sequences	Off-target reads or artifacts	Contact Element Technical Support

For more information on assignment thresholds and additional troubleshooting, see Cell assignment logic.

3′ Transcriptome

For each 3′ Transcriptome batch, Cells2Stats writes {root}/Wells/{well}/{batch}/{tile}_rawreads.parquet. Each file contains one row per polony, with the following columns:

Column	Description
ReadID	Read identifier. See Read identifiers.
PolonyIndex	Polony index within the tile
X, Y	Polony position in the tile, in pixels
Xum, Yum, Zum	Polony position, in micrometers
WellX, WellY, WellXum, WellYum	Polony position in the well, in pixels and micrometers
Cell	ID of the cell that contains the polony
IsNuclear	Whether the polony is inside the nucleus

Barcoding

For each barcoding batch, Cells2Stats writes {root}/Wells/{well}/{batch}/{tile}_barcodes.parquet. Each file contains one row per polony, with the following columns:

Column	Description
ReadID	Read identifier. See Read identifiers.
PolonyIndex	Polony index within the tile
X, Y	Polony position in the tile, in pixels
Xum, Yum, Zum	Polony position, in micrometers
WellX, WellY, WellXum, WellYum	Polony position in the well, in pixels and micrometers
BarcodeIndex	Index of the barcode assigned to the polony
TargetName	Target that the polony is assigned to
Cell	ID of the cell that contains the polony
IsNuclear	Whether the polony is inside the nucleus

QC report

Cells2Stats generates qc_report.html, an interactive HTML quality control report for the run, in the root of the output directory. The report summarizes run-level and well-level quality, per-target intensity and segmentation metrics, and detected spatial artifacts, and includes a flow cell summary image. The report includes a tab for each assay in the run.

Versions

The Versions.json reports the version number for CSV output files and bundled software programs. The following table describes the fields:

Field	Description	Data Type
FileVersion	Overall version of the file	String
FileVersions	Versions of individual files in the dataset	Object
AverageNormWellStats.csv	Version of AverageNormWellStats.csv file	String
RawCellStats.csv	Version of RawCellStats.csv file	String
ProgramVersions	Versions of programs that are used in the analysis	Object
CellProfiler	Version of the CellProfiler program	String

Segmentation mask formats

Cells2Stats copies segmentation masks from the run directory without converting them, so the mask format follows the AVITI24™ software version that produced the run:

AVITI OS version	Cell mask	Nuclear mask
4.1.0 and later	{tile}_Cell.zarr, Zarr v3, uint32	{tile}_Nuclear.zarr, Zarr v3, uint8
Earlier than 4.1.0	{tile}_Cell.tif	{tile}_Nuclear.tif

Cells2Stats reads both formats, so a run in either format analyzes without additional configuration. For the input-side detail, see Segmentation mask formats. Downstream tools that read mask files directly must handle whichever format the run carries.

Caution:

A tool that reads only TIFF masks fails on a run produced by AVITI OS 4.1.0 or later. CellProfiler requires uint16 TIFF input, so a Zarr cell mask must be chunked and converted before you use it for re-segmentation. For the full procedure, see Evaluating and applying alternate segmentation models.

Read identifiers

Cell screening read identifiers carry a trailing polony index, a 7-digit integer that resets for each z-tile. In the following example, 0002538 is the polony index:

@…:0:B01:0002538

Other files

Some files are copied from the run directory, such as RunManifest.json, RunParameters.json, Panel.json, cell segmentation masks, and nuclear segmentation masks. For more information on these files, see Cell Screening Run Output Files.