Release notes

Version 4

4.0.4 Apr 14, 2026

  • Cellranger workflow:

    • Support Flex v2 data processing. Users need to specify flex-v2 in DataType column of the input sample sheet. flex-v1 and frp are for Flex v1.

  • SmartSeq2 workflow:

    • Default resources are stored in gs://cumulus-ref bucket in GCP region us-central1.

4.0.3 Apr 6, 2026

  • Cellranger workflow:

    • Fix a bug in processing 10x On-Chip Multiplexing (OCM) data.

    • Support input FASTQ files in unzipped .fastq format.

  • Demultiplexing workflow:

    • For Souporcell, add souporcell_umi_tag to allow users specify a custom UMI tag other than the default UB for their data.

4.0.2 Jan 6, 2026

  • Cellranger workflow:

    • Bug fix on processing VDJ data.

4.0.1 Dec 14, 2025

  • Cellranger workflow:

    • For Flex data, avoid using genome reference when no_bam = true and cellranger_version >= "8.0.0".

4.0.0 Nov 26, 2025

  • Cellranger workflow:

    • Support library-level chemistry for Flex data processing. This is needed when Flex and CITE-Seq libraries have probe barcodes at different read pairs, which causes the auto-detection to fail (e.g. MFRP-Ab-R1 for CITE-Seq library, MFRP-RNA for Flex library). (PR 446)

    • Upgrade cellranger_version default to 10.0.0.

    • Upgrade cellranger_arc_version* default to 2.1.0, with associated important feature changes:

      • The 20,000 total cell limit is removed.

      • Add secondary input for running secondary analysis or not, with default false.

    • Upgrade cellranger_atatc_version default to 2.2.0, with associated important feature changes:

      • The 20,000 total cell limit is removed.

      • The force_cells input can be any positive integer, which is no longer restricted to be smaller than 20,000.

      • Add secondary input for running secondary analysis or not, with default false.

    • Add 2024-A transcriptome references for Cell Ranger ARC and ATAC: GRCh38-2024-A_arc and GRCm39-2024-A_arc.

  • Spaceranger workflow:

    • Upgrade spaceranger_version default to 4.0.1, with associated important feature changes:

      • Support Visium HD 3’. To enable it, in the sample sheet, leave the ProbeSet column blank for such samples, or skip this column if all samples are Visium HD 3’.

      • Add Visium probesets v2.1 which are bundled with 2024-A transcriptome references: human_probe_v2.1 and mouse_probe_v2.1.

Version 3

3.1.1 May 22, 2025

  • Demultiplexing workflow:

    • Upgrade demuxEM_version default to 0.1.8, which supports HTO and CRISPR count matrices in 10x hdf5 format.

    • Resolve numpy out-of-memory-map issue when running on large AWS EC2 instances. (PR 439 thanks to Dantistnfs)

3.1.0 May 12, 2025

  • Cellranger workflow:

    • Upgrade cumulus_feature_barcoding_version default to 2.0.0.

      • Apply UMI correction to raw count matrices; for crispr type samples, further apply PCR chimeric filtering.

    • Rename min_read_ratio to read_ratio_cutoff, and default is changed from 0.1 to 0.5. Still only applied to crispr type samples.

    • Samples of crispr, hashing, citeseq, adt and cmo types now accept Reference column values.

3.0.0 March 7, 2025

  • Overall highlights:

    • For data localization/delocalization on GCP, now use gcloud storage commands instead of gsutil, as gsutil is deprecated, and gcloud storage achieves some speed improvement.

    • Users no longer need to specify backend input for AWS or local backend, and it is now automatically figured out from output_directory location.

    • Remove support on mkfastq in Cellranger and Spaceranger workflows, as it will soon be removed from Cell Ranger and Space Ranger. Users need to run BCL Convert themselves first to generate FASTQ data.

    • For Cellranger and Spaceranger workflows, provide better support for shared computing environments like AWS Batch, GCP Batch and sHPC.

    • For resources like prebuilt references, they are now held in a single-region bucket gs://cumulus-ref in US-CENTRAL1 to reduce potentially higher network cost at users’ side from the previous multi-region bucket.

  • Cellranger workflow:

    • Upgrade cellranger_version default to 9.0.1.

    • Remove mkfastq related workflow inputs.

    • Remove run_count input as it’s always true.

    • Support input data format as a TAR file containing FASTQ files.

    • Change FeatureBarcodeFile column header to AuxFile (for backward compatibility, FeatureBarcodeFile is still accepted but not recommended).

    • Support all the 3 Sample Multiplexing methods provided since Cell Ranger v9.0. See details

    • For single-cell and single-nucleus RNA-seq:

      • Add new genome reference mRatBN7.2-2024-A

      • Remove Target Gene Expression-related inputs, as it’s no longer supported since Cell Ranger v7.2.0.

    • For feature barcoding:

      • Upgrade cumulus_feature_barcoding_version to 1.0.0.

      • The workflow now can automatically detect the chemistry type of the data:

        • auto by default: The workflow checks all possible assay types to decide the correct one.

        • threeprime: The workflow checks all 3’ assay types to decide the correct one.

        • fiveprime: The workflow checks all 5’ assay types to decide the correct one.

      • Reorg the keywords in Chemistry column of sample sheet: 5’ now only has 2 types, SC5Pv2 and SC5Pv3 for 5’ v2 and v3 chemistries where only R2 is used for alignment.

      • Support the new format of 10x cell barcode inclusion lists provided in Cell Ranger v9.0+.

      • Fix issue in processing UTF-encoded feature barcode files

    • For immune profiling:

      • Add types vdj_t, vdj_b and vdj_t_gd for DataType column. See 10x 5’ Immune Profiling Kit for details.

      • Remove chain input, as it is now automatically decided by user-specified DataType types.

      • For vdj_t_gd type samples, support the feature of specifying primer sequences used to enrich cDNA for V(D)J sequences. To enable it, provide a .txt file in AuxFile column of the sample, and it will be passed to --inner-enrichment-primers option of cellranger vdj in execution.

    • For Flex Gene Expression:

      • Remove ProbeSet column. The probe set is now automatically decided based on user-specified Reference name.

      • Support Flex probe sets v1.1 which are associated with 2024-A genome references.

    • For CellPlex using CMO:

      • Remove cmo_set input. If using custom CMOs in your experiment, just provide the custom feature reference file in AuxFile column of the cmo type sample.

  • Spaceranger workflow:

    • Upgrade spaceranger_version default to 3.1.3.

    • Remove mkfastq related workflow inputs.

    • Remove run_count input as it’s always true.

    • Remove support on Targeted Gene Expression analysis, as it’s no longer supported since Space Ranger v2.1.1.

Version 2.6

2.6.3 August 2, 2024

  • Update Demultiplexing workflow to work with Stratocumulus v0.2.4.

2.6.2 June 19, 2024

  • Solve the issue of Cellranger workflow with cellranger-arc. cellranger_arc_version default is now “2.0.2.strato” which is compatible with workflow v2.6.1 or later.

2.6.1 May 8, 2024

  • In Cellranger workflow: Add new genome references for single-cell/nucleus RNA-Seq: GRCh38-2024-A for human, GRCm39-2024-A for mouse, and GRCh38_and_GRCm39-2024-A for human and mouse.

  • In Spaceranger workflow: Add a new probe set mouse_probe_v2 for mouse.

  • Some underlying workflow improvement.

2.6.0 April 22, 2024

Update:

  • In Cellranger workflow:

    • Upgrade cellranger_version default to 8.0.0.

    • Upgrade cumulus_feature_barcoding version default to 0.11.3.

  • In Spaceranger workflow:

    • Upgrade spaceranger_version default to 3.0.0.

    • Support Visium HD data.

Version 2.5

2.5.0 February 3, 2024

Improvements:

  • In Cellranger workflow:

    • Add multi_disk_space input for specifying disk size used by Fixed RNA Profiling or 10x multiome jobs, which usually take large amount.

  • In Cellbender workflow:

    • With 0.3.0 default Cell Bender version, users don’t need to always specify expected_cells or total_droplets_included input.

Updates:

  • In Cellranger workflow:

    • Upgrade cellranger_version default to 7.2.0.

    • Upgrade cumulus_feature_barcoding version default to 0.11.2.

    • Add GRCh38 VDJ v7.1.0 reference: GRCh38_vdj_v7.1.0.

  • In Spaceranger workflow:

    • Upgrade spaceranger_version default to 2.1.1.

  • In Demultiplexing workflow:

    • Upgrade souporcell_version default to 2.5.

  • In Cellbender workflow:

    • Upgrade cellbender_version default to 0.3.0.

Version 2.4

2.4.1 May 30, 2023

Improve:

  • In Cellranger workflow:

    • Fixed RNA Profiling now accepts custom probe set references.

  • In STARsolo workflow:

    • Add limitBAMsortRAM and outBAMsortingBinsN inputs to handle out-of-memory error in BAM sorting phase.

  • In GeoMx_fastq_to_dcc workflow:

    • Support multiple FASTQ folders as input.

Updates:

  • In Demultiplexing workflow:

    • Upgrade souporcell_version to 2022.12, which is based on commit 9fb527 on 2022/12/13.

  • In STARsolo workflow:

    • Upgrade star_version to 2.7.10b.

Bug Fixes:

  • In Spaceranger workflow:

    • Fix the image localization issue for CytAssist samples.

2.4.0 January 28, 2023

Updates:

  • In Cellranger workflow:

    • Upgrade cellranger_version default to 7.1.0.

    • Add Mouse probe set v1.0 for Fixed RNA Profiling analysis.

    • Add probe set v1.0.1 of both Human and Mouse for Fixed RNA Profiling analysis.

    • Upgrade cumulus_feature_barcoding version default to 0.11.1.

  • In Cellranger_create_reference workflow: upgrade cellranger_version default to 7.1.0.

  • In Cellranger_vdj_create_reference workflow: upgrade cellranger_version default to 7.1.0.

  • In Spaceranger workflow: upgrade spaceranger_version default to 2.0.1.

Version 2.3

2.3.0 October 30, 2022

New Features:

Updates:

  • In Cellranger workflow:

    • Upgrade cellranger_version default to 7.0.1.

    • Upgrade cellranger_arc_version default to 2.0.2.

  • In Cellranger_create_reference workflow: upgrade cellranger_version default to 7.0.1.

  • In Cellranger_vdj_create_reference workflow: upgrade cellranger_version default to 7.0.1.

Version 2.2

2.2.0 October 4, 2022

New Features:

  • Add Nanostring GeoMx DSP workflows. It consists of two steps:

    • GeoMx_fastq_to_dcc workflow to convert FASTQ files into DCC files by wrapping Nanostring GeoMx Digital Spatial NGS Pipeline.

    • GeoMx_dcc_to_count_matrix workflow to generate probe count matrix from DCC files with pathologists’ annotation.

Updates:

  • Spaceranger workflow:

    • Add support on 10x Space Ranger v2.0.0.

    • Add human_probe_v2 Probe Set for FFPE samples, which is compatible with CytAssist FFPE samples.

  • Upgrade cumulus_feature_barcoding_version default to v0.11.0 for Feature Barcoding in Cellranger workflow.

API Changes:

  • Across all workflows, for AWS backend:

    • All workflows now have awsQueueArn input, which is used for explicitly specifying the Arn string of an AWS Compute Environment.

    • Remove awsMaxRetries input for all workflows. Namely, use Cromwell’s default value 0.

Bug Fix:

  • Fix the issue on localizing GCP folders in Cellranger workflow for ATAC-Seq and 10x Multiome data.

Version 2.1

2.1.1 July 18, 2022

  • Make cumulus workflow work with Cromwell v81+.

2.1.0 July 13, 2022

New Features:

  • Add CellBender workflow for ambient RNA removal.

  • CellRanger:

    • For ATAC-Seq data, add ARC-v1 chemistry keyword for analyzing only the ATAC part of 10x multiome data. See CellRanger scATAC-seq sample sheet section for details.

    • For antibody/hashing/citeseq/crispr data, add multiome chemistry keyword for the feature barcoding on 10x multiome data.

  • STARsolo:

    • In workflow output, besides mtx format gene-count matrices, the workflow also generates matrices in 10x-compatible hdf5 format.

Improvements:

  • CellRanger: For antibody/hashing/citeseq/crispr data,

    • cumulus_feature_barcoding v0.9.0+ now supports multi-threading and faster gzip file I/O.

  • Workflows check if output_directory is a valid Cloud URI based on the given backend value before execution. (Feature request #322 )

Updates:

  • Genome Reference:

  • Default version upgrade:

    • Update cellranger default to v7.0.0.

    • Update cellranger-atac default to v2.1.0.

    • Update cellranger-arc default to v2.0.1.

    • Update cumulus_feature_barcoding default to v0.10.0.

    • STARsolo workflow uses STAR v2.7.10a by default.

Version 2.0

2.0.0 March 14, 2022

Overall:

Workflow-specific:

  • Add STARsolo_create_reference workflow to build genome references for STARsolo counting. See its documentation for details.

  • On Cellranger workflow:

    • Add support for 10x Cell Ranger version 6.1.1 and 6.1.2, and use 6.1.2 by default. See Cell Ranger v6.1 release notes.

    • Add support for 10x Cell Ranger ARC version 2.0.1, and use it by default. See Cell Ranger ARC v2.0 release notes for the release notes.

    • Upgrade cumulus_feature_barcoding to version 0.7.0 to allow manually set barcode starting position (via input crispr_barcode_pos).

    • Add support for non 10x CRISPR assays. See the description of crispr DataType value in this section for details.

    • For input data consisting of fastq files, it’s able to handle folder structure of both flat (all fastq files in one folder) and nested (one subfolder per sample listed in the input sample sheet) forms.

    • Add fastq_outputs to workflow output, which contains mkfastq step output folders for samples listed in the input sample sheet.

    • Add count_outputs to workflow output, which contains count step output folderrs for samples listed in the input sample sheet.

  • On Spaceranger workflow:

    • Add support for 10x Space Ranger version 1.3.0 and 1.3.1, and use 1.3.1 by default. See Space Ranger v1.3 release notes for the release notes.

    • For input data consisting of fastq files, it’s able to handle folder structure of both flat (all fastq files in one folder) and nested (one subfolder per library) forms.

    • Add output section for the workflow. See here for details.

    • Retire old genome references:

      • Keep GRCh38-2020-A and mm10-2020-A.

      • Retire GRCh38, mm10, GRCh38-2020-A-premrna and mm10-2020-A-premrna. Users can still reach out to Cumulus team to ask for URIs to these old references, but they are not provided by default.

    • In the description of ReorientImages field of input sample sheet, add the information on its valid values.

  • On STARsolo workflow:

    • Add support for STAR version 2.7.9a, and use it by default. See STAR v2.7.9a release notes for the release notes.

    • Reorganize the workflow by exposing more inputs to users.

    • Add support on more protocols: 10x multiome, 10x 5’ (both SC5P-R2 and SC5P-PE), Slide-Seq and Share-Seq. See here <./starsolo.html#prepare-a-sample-sheet> for details.

    • Use input read1_fastq_pattern and read2_fastq_pattern to support fastq files generated by Cell Ranger or SeqWell, as well as Sequence Read Archive (SRA) data.

    • For input data consisting of fastq files, it’s able to handle folder structure of both flat (all fastq files in one folder) and nested (one subfolder per library) forms.

    • Do not attach filename prefix to output files to avoid the incorrect SJ raw feature.tsv symlink error, which would cause the folder delocalization fail. (see discussion with STAR team)

    • Add STAR log file to workflow output. This is the Log.out file if running STAR locally, which can be used for tracking the process and sharing with STAR team when opening an issue there.

    • Retire old genome references:

      • Keep GRCh38-2020-A, mm10-2020-A, and GRCh38-and-mm10-2020-A.

      • Retire old references listed here. Users can still reach out to Cumulus team to ask for URIs to them, but they are not provided by default.

  • On Demultiplexing workflow:

    • Upgrade demuxEM to version 0.1.7 for bug fix.

  • On Cellranger_create_reference workflow:

    • Add the generated reference file to the workflow output.

    • Bug fix in using input memory.

    • Update documentation to suggest only using Cell Ranger version 6.1.1 or later for building reference, as v6.0.1 has issues which leave the job running without terminating.

  • On Cellranger_atac_create_reference workflow:

    • Add the generated reference file to the workflow output.

  • On Cellranger_vdj_create_reference workflow:

    • Add the generated reference file to the workflow output.

Version 1.x

Version 1.5.1 September 15, 2021

  • Fix the issue of WDLs after Terra platform updates the Cromwell engine.

Version 1.5.0 July 20, 2021

  • On demultiplexing workflow

    • Update demuxEM to v0.1.6.

  • On cumulus workflow

    • Add Nonnegative Matrix Factorization (NMF) feature: run_nmf and nmf_n inputs.

    • Add integrative NMF (iNMF) data integration method: inmf option in correction_method input; the number of expected factors is also specified by nmf_n input.

    • When NMF or iNMF is enabled, word cloud plots and gene program UMAP plots of NMF/iNMF results will be generated.

    • Update Pegasus to v1.4.2.

Version 1.4.0 May 17, 2021

  • On cellranger workflow

    • Add support for multiomics analysis using linked samples, cellranger-arc count, cellranger multi and cellranger count will be automatically triggered based on the sample sheet

    • Add support for cellranger version 6.0.1 and 6.0.0

    • Add support for cellranger-arc version 2.0.0, 1.0.1, 1.0.0

    • Add support for cellranger-atac version 2.0.0

    • Add support for cumulus_feature_barcoding version 0.6.0, which handles CellPlex CMO tags

    • Add GRCh38-2020-A_arc_v2.0.0, mm10-2020-A_arc_v2.0.0, GRCh38-2020-A_arc_v1.0.0 and mm10-2020-A_arc_v1.0.0 references for cellranger-arc.

    • Fixed bugs in cellranger_atac_create_reference

    • Add delete undetermined FASTQs option for mkfastq

  • On demultiplexing workflow

    • Replace demuxlet with popscle, which includes both demuxlet and freemuxlet

  • On cumulus workflow

    • Fixed bug that remap_singlets and subset_singlets don’t work when input is in sample sheet format.

  • Modified workflows to remove trailing spaces and support spaces within output_directory

Version 1.3.0 February 2, 2021

  • On cumulus workflow:

    • Change cumulus_version to pegasus_version to avoid confusion.

    • Update to use Pegasus v1.3.0 for analysis.

Version 1.2.0 January 19, 2021

  • Add spaceranger workflow:

    • Wrap up spaceranger version 1.2.1

  • On cellranger workflow:

    • Fix workflow WDL to support both single index and dual index

    • Add support for cellranger version 5.0.1 and 5.0.0

    • Add support for targeted gene expression analysis

    • Add support for --include-introns and --no-bam options for cellranger count

    • Remove --force-cells option for cellranger vdj as noted in cellranger 5.0.0 release note

    • Add GRCh38_vdj_v5.0.0 and GRCm38_vdj_v5.0.0 references

  • Bug fix on cumulus workflow.

  • Reorganize the sidebar of Cumulus documentation website.

Version 1.1.0 December 28, 2020

  • On cumulus workflow:

    • Add CITE-Seq data analysis back. (See section Run CITE-Seq analysis for details)

    • Add doublet detection. (See infer_doublets, expected_doublet_rate, and doublet_cluster_attribute input fields)

    • For tSNE visualization, only support FIt-SNE algorithm. (see run_tsne and plot_tsne input fields)

    • Improve efficiency on log-normalization and DE tests.

    • Support multiple marker JSON files used in cell type annotation. (see organism input field)

    • More preset gene sets provided in gene score calculation. (see calc_signature_scores input field)

  • Add star_solo workflow (see STARsolo section for details):

    • Use STARsolo to generate count matrices from FASTQ files.

    • Support chemistry protocols such as 10X-V3, 10X-V2, DropSeq, and SeqWell.

  • Update the example of analyzing hashing and CITE-Seq data (see Example section) with the new workflows.

  • Bug fix.

Version 1.0.0 September 23, 2020

  • Add demultiplexing workflow for cell-hashing/nucleus-hashing/genetic-pooling analysis.

  • Add support on CellRanger version 4.0.0.

  • Update cumulus workflow with Pegasus version 1.0.0:

    • Use zarr file format to handle data, which has a better I/O performance in general.

    • Support focus analysis on Unimodal data, and appending other Unimodal data to it. (focus and append inputs in cluster step).

    • Quality-Control: Change percent_mito default from 10.0 to 20.0; by default remove bounds on UMIs (min_umis and max_umis inputs in cluster step).

    • Quality-Control: Automatically figure out name prefix of mitochondrial genes for GRCh38 and mm10 genome reference data.

    • Support signature / gene module score calculation. (calc_signature_scores input in cluster step)

    • Add Scanorama method to batch correction. (correction_method input in cluster step).

    • Calculate UMAP embedding by default, instead of FIt-SNE.

    • Differential Expression (DE) analysis: remove inputs mwu and auc as they are calculated by default. And cell-type annotation uses MWU test result by default.

  • Remove cumulus_subcluster workflow.

Version 0.x

Version 0.15.0 May 6, 2020

  • Update all workflows to OpenWDL version 1.0.

  • Cumulus now supports multi-job execution from Terra data table input.

  • Cumulus generates Cirrocumulus input in .cirro folder, instead of a huge .parquet file.

Version 0.14.0 February 28, 2020

  • Added support for gene-count matrices generation using alternative tools (STARsolo, Optimus, Salmon alevin, Kallisto BUStools).

  • Cumulus can process demultiplexed data with remapped singlets names and subset of singlets.

  • Update VDJ related inputs in Cellranger workflow.

  • SMART-Seq2 and Count workflows are in OpenWDL version 1.0.

Version 0.13.0 February 7, 2020

  • Added support for aggregating scATAC-seq samples.

  • Cumulus now accepts mtx format input.

Version 0.12.0 December 14, 2019

  • Added support for building references for sc/snRNA-seq, scATAC-seq, single-cell immune profiling, and SMART-Seq2 data.

Version 0.11.0 December 4, 2019

  • Reorganized Cumulus documentation.

Version 0.10.0 October 2, 2019

  • scCloud is renamed to Cumulus.

  • Cumulus can accept either a sample sheet or a single file.

Version 0.7.0 Feburary 14, 2019

  • Added support for 10x genomics scATAC assays.

  • scCloud runs FIt-SNE as default.

Version 0.6.0 January 31, 2019

  • Added support for 10x genomics V3 chemistry.

  • Added support for extracting feature matrix for Perturb-Seq data.

  • Added R script to convert output_name.seurat.h5ad to Seurat object. Now the raw.data slot stores filtered raw counts.

  • Added min_umis and max_umis to filter cells based on UMI counts.

  • Added QC plots and improved filtration spreadsheet.

  • Added support for plotting UMAP and FLE.

  • Now users can upload their JSON file to annotate cell types.

  • Improved documentation.

  • Added lightGBM based marker detection.

Version 0.5.0 November 18, 2018

  • Added support for plated-based SMART-Seq2 scRNA-Seq data.

Version 0.4.0 October 26, 2018

  • Added CITE-Seq module for analyzing CITE-Seq data.

Version 0.3.0 October 24, 2018

  • Added the demuxEM module for demultiplexing cell-hashing/nuclei-hashing data.

Version 0.2.0 October 19, 2018

  • Added support for V(D)J and CITE-Seq/cell-hashing/nuclei-hashing.

Version 0.1.0 July 27, 2018

  • KCO tools released!