Feature Reference CSV Filling Guide for Antibody Capture, CRISPR Guide Capture, and Cell Multiplexing

Author: SeekGene

Time: 17 min

Words: 3.3k words

Updated: 2026-07-31

Reads: 0 times

This document helps you fill in Feature Barcode information (Antibody, CRISPR, Hashing, etc.) for Cell Ranger analysis. Please select the appropriate content based on the product type and library construction method (3' / 5').

General Notes for Filling the Reference Table

Product Overview

Product	Description	Target
Antibody Capture (Cell Surface Protein / CITE-seq)	Simultaneously quantify surface proteins in single-cell transcriptome sequencing using antibody-oligonucleotide tags	Surface proteins (e.g., CD3, CD4)
CRISPR Guide Capture (CRISPR screening / Perturb-seq)	Capture sgRNAs in individual cells to link genetic perturbations with transcriptomic phenotypes (see "CRISPR Guide Capture" section)	sgRNA sequences
Cell Multiplexing (Hashing / CellPlex CMO)	Pool multiple samples for sequencing and computationally demultiplex them into individual samples using barcode tags	Sample origin information

Basic Concepts (Common to All Products)

Whether for Antibody Capture, CRISPR Guide Capture, or Cell Multiplexing, Cell Ranger requires a Feature Reference CSV file that tells the software:

The name and ID of each feature (antibody / sgRNA / Hashtag).
The barcode sequence of each feature.
The exact position and extraction pattern of the barcode sequence within the sequencing read.
The type of analysis each feature belongs to (antibody, CRISPR, or multiplexing).

All product types share 6 common required columns with fixed column names and order. CRISPR products require 2 additional columns (see "CRISPR Guide Capture" section):

Column	Description	Required
`id`	Unique feature ID (e.g., CD3, sgTP53-1, Hashtag1). Must not contain spaces, slashes, quotes, commas, or other special characters. Must not conflict with transcriptome gene names.	✅ Required
`name`	Feature display name. Can be the same as `id`. Uniqueness not required. Displayed in `matrix/features.tsv.gz`. Same special character restrictions apply.	✅ Required
`read`	Sequencing read containing the feature barcode. Typically `R2` (rarely `R1`).	✅ Required
`pattern`	Fixed sequence indicating the barcode position within the read (see details below).	✅ Required
`sequence`	Barcode sequence. For example, the antibody oligonucleotide sequence or the sgRNA protospacer sequence.	✅ Required
`feature_type`	Feature type: `Antibody Capture` / `CRISPR Guide Capture` / `Multiplexing Capture`.	✅ Required

Pattern Syntax

Symbol	Meaning
`5P` or `^`	Indicates that the pattern starts matching from the beginning of the read. Can only appear at the start of the pattern.
`3P` or `$`	Indicates that the pattern is at the end of the read. Can only appear at the end.
`N`	Any nucleotide (placeholder, not validated).
`A` / `C` / `G` / `T`	Fixed nucleotide that must match the `sequence` column exactly.
`(BC)`	Marks the position of the actual Feature Barcode sequence. Must appear exactly once per pattern. Corresponds to the `sequence` column.

Pattern Examples:

5PNNNNNNNNNN(BC): The pattern starts at the beginning of the read, and the sequence immediately after it is the sequence.
^(BC): The sequence starts directly at the beginning of the read.
(BC)GTTTAAGAGCTAAGCTGGAA: The pattern sequence is GTTTAAGAGCTAAGCTGGAA, and the sequence is located before it.
TTCCAGCTTAGCTCTTAAAC(BC): The pattern sequence is TTCCAGCTTAGCTCTTAAAC, and the sequence is located after it.
The pattern length should be moderate: too long may be affected by sequencing errors, too short may not uniquely identify the sequence. For non-N fixed bases, a length of 12-20 bp is recommended.
Cell Ranger identifies and extracts the sequence based on the pattern, then matches it against the reference table. Up to one base mismatch is allowed, and a barcode correction algorithm is applied.

General Filling Guidelines (Must Check)

The file must be in pure ASCII/UTF-8 CSV format. Do not use Excel-saved files with full-width Chinese punctuation, BOM headers, or special characters.
Each row corresponds to a unique Feature Barcode. The id column must be unique across the entire file (including both antibodies and multiplexing tags in the same experiment) and must not conflict with gene names or probe IDs.
The id column should only contain letters, numbers, underscores _, and hyphens -. Do not use pipes |, commas, or spaces. Pipes have special meaning in multi config CSV.
If an experiment includes both antibody detection and sample multiplexing (Hashing), both feature sets must be written in the same Feature Reference CSV file (with different feature_type values). Do not split them into two files.
The sequence column must be verified as correct. For sequences sourced from vendors, cross-check against the latest sequence table provided by the antibody/sgRNA/Hashtag supplier. For CRISPR products, ensure the provided sequence is the sgRNA strand that is actually transcribed and sequenced. Incorrect sequences will result in the feature not being detected (zero counts).

Antibody Capture

Antibody Capture assays typically use BioLegend TotalSeq™ series. The position of the barcode sequence within the R2 read varies by antibody type. Sequencing data filenames usually end with _Pro. The feature_type is always Antibody Capture.

TotalSeq-A (Compatible with 3' v2 / v3 kits)

Compatible data type: 3' gene expression (Single Cell 3' v2 / v3).
The barcode sequence starts at the very beginning of the R2 read (no preceding spacer sequence).
read = R2, pattern = 5P(BC).

Example (only modify id / name / sequence according to your actual experiment):

csv

id,name,read,pattern,sequence,feature_type
CD3,CD3_TotalSeqA,R2,5P(BC),ATTGGCACTCAGATG,Antibody Capture
CD4,CD4_TotalSeqA,R2,5P(BC),TACCCGTAATAGCGT,Antibody Capture

Download template

TotalSeq-B (Compatible with 3')

Compatible data type: 3' gene expression.
The barcode sequence starts at the 10th base of the R2 read (preceded by a 10 bp random sequence).
read = R2, pattern = 5PNNNNNNNNNN(BC).

Example (only modify id / name / sequence according to your actual experiment):

csv

id,name,read,pattern,sequence,feature_type
CD3,CD3_TotalSeqB,R2,5PNNNNNNNNNN(BC),AACAAGACCCTTGAG,Antibody Capture
CD4,CD4_TotalSeqB,R2,5PNNNNNNNNNN(BC),TACCCGTAATAGCGT,Antibody Capture

Download template

TotalSeq-C (Compatible with 5' / V(D)J)

Compatible data type: 5' gene expression / V(D)J immune profiling.
The barcode sequence also starts at the 10th base of the R2 read.
read = R2, pattern = 5PNNNNNNNNNN(BC).

Example (only modify id / name / sequence according to your actual experiment):

csv

id,name,read,pattern,sequence,feature_type
CD3,CD3_TotalSeqC,R2,5PNNNNNNNNNN(BC),CTCATTGTAACTCCT,Antibody Capture
CD19,CD19_TotalSeqC,R2,5PNNNNNNNNNN(BC),CTGGGCAATTACTCG,Antibody Capture

Download template

Cell Multiplexing

Cell multiplexing pools multiple samples for sequencing and computationally demultiplexes them using barcode tags (Hashtag / CMO). 10x Genomics supports the following multiplexing methods. Please confirm which one you are using:

Multiplexing Method	Platform	Tag Source	Feature Reference Required?
3' CellPlex (CMO)	3' gene expression	10x official CMO tags (built-in, usually no custom reference needed)	Not required when using official CMOs; required when using third-party/custom tags (e.g., TotalSeq for Hashing)
Antibody Capture Hashing (TotalSeq Hashtag)	3' or 5' gene expression	Hashtag sequences from antibody vendors	✅ Required

Currently, 3' multiplexing data filenames usually end with _HTO or _ADT. 5' multiplexing data filenames end with _Pro.

Antibody Capture Hashing (Most Common, Surface Protein-Based Multiplexing)

When using TotalSeq-A / B / C Hashtag antibodies for sample multiplexing:

Provide a custom Feature Reference. The format is identical to that of the corresponding TotalSeq antibody type (e.g., TotalSeq-B uses 5PNNNNNNNNNN(BC), TotalSeq-C uses 5PNNNNNNNNNN(BC)).
Inform us in advance that this is a surface protein-based multiplexing experiment for proper bioinformatics processing.
Set feature_type to Antibody Capture.
Confirm whether you are using CellPlex CMO official tags or TotalSeq Hashtag antibodies. The feature_type values differ (Multiplexing Capture vs. Antibody Capture), and mixing them up will cause analysis failure.
Multiplexing analysis requires Gene Expression data to be provided simultaneously.

Multiplexing with Surface Protein Hashtags Only

Example with TotalSeq-A Hashtag (for 3' data; only modify id / name / sequence according to your actual experiment):

csv

id,name,read,pattern,sequence,feature_type
TotalSeqA_Hashtag_1,TotalSeqA_Hashtag_1,R2,5P(BC),ATCACATCGTTGCCA,Antibody Capture
TotalSeqA_Hashtag_2,TotalSeqA_Hashtag_2,R2,5P(BC),CTCTAGGTTCCCACC,Antibody Capture
TotalSeqA_Hashtag_3,TotalSeqA_Hashtag_3,R2,5P(BC),AGCTCACTTGTCCTG,Antibody Capture
TotalSeqA_Hashtag_4,TotalSeqA_Hashtag_4,R2,5P(BC),GTATGTTCTGCAATT,Antibody Capture

Download template

Example with TotalSeq-B Hashtag (for 3' data; only modify id / name / sequence according to your actual experiment):

csv

id,name,read,pattern,sequence,feature_type
TotalSeqB_Hashtag_1,TotalSeqB_Hashtag_1,R2,5PNNNNNNNNNN(BC),GTCAACTCTTTAGCG,Antibody Capture
TotalSeqB_Hashtag_2,TotalSeqB_Hashtag_2,R2,5PNNNNNNNNNN(BC),TGATGGCCTATTGGG,Antibody Capture

Download template

Example with TotalSeq-C Hashtag (for 5' data; only modify id / name / sequence according to your actual experiment):

csv

id,name,read,pattern,sequence,feature_type
TotalSeqC_Hashtag_1,TotalSeqC_Hashtag_1,R2,5PNNNNNNNNNN(BC),GTCAACTCTTTAGCG,Antibody Capture
TotalSeqC_Hashtag_2,TotalSeqC_Hashtag_2,R2,5PNNNNNNNNNN(BC),TGATGGCCTATTGGG,Antibody Capture
TotalSeqC_Hashtag_3,TotalSeqC_Hashtag_3,R2,5PNNNNNNNNNN(BC),TTCCGCCTCTCTTTG,Antibody Capture
TotalSeqC_Hashtag_4,TotalSeqC_Hashtag_4,R2,5PNNNNNNNNNN(BC),AGTAAGTTCAGCGTA,Antibody Capture

Download template

Both Antibody Capture and Hashtag Multiplexing

Example using TotalSeq-B tags (modify id / name / sequence according to your actual experiment):

csv

id,name,read,pattern,sequence,feature_type
TotalSeqB_Hashtag_1,TotalSeqB_Hashtag_1,R2,5PNNNNNNNNNN(BC),GTCAACTCTTTAGCG,Antibody Capture
TotalSeqB_Hashtag_2,TotalSeqB_Hashtag_2,R2,5PNNNNNNNNNN(BC),TGATGGCCTATTGGG,Antibody Capture
CD3_TotalSeqB,CD3,R2,5PNNNNNNNNNN(BC),AACAAGACCCTTGAG,Antibody Capture
CD4_TotalSeqB,CD4,R2,5PNNNNNNNNNN(BC),TACCCGTAATAGCGT,Antibody Capture
CD8a_TotalSeqB,CD8a,R2,5PNNNNNNNNNN(BC),ATTGGCACTCAGATG,Antibody Capture

Download template

Bioinformatics Notes

During analysis, the bioinformatician will add sample_id and hashtag_ids entries in the [samples] section of the configuration file for sample demultiplexing (handled by bioinformatics).

The demultiplexed sample name, sample_id, and hashtag_ids must all match the id values.

csv

[samples]
sample_id,hashtag_ids
TotalSeqB_Hashtag_1,TotalSeqB_Hashtag_1
TotalSeqB_Hashtag_2,TotalSeqB_Hashtag_2

If the same sample uses multiple Hashtag antibodies (technical replicates), separate the IDs with a pipe |, e.g., ABHT-1|ABHT-2. Example (handled by bioinformatics):

WARNING

Please inform us in advance about such cases!

csv

[samples]
sample_id,hashtag_ids
Sample1,TotalSeqB_Hashtag_1|TotalSeqB_Hashtag_2
Sample2,TotalSeqB_Hashtag_3

3' CellPlex with CMO

CMO (Cell Multiplexing Oligo) uses lipid-anchored oligonucleotides for multiplexing.

Multiplexing analysis requires Gene Expression data to be provided simultaneously.
Set feature_type to Multiplexing Capture.
Confirm whether the CMO tags are the default Cell Ranger ones or custom.

Using 10x Official CMO Tags

10x provides 12 default CMO tags (CMO301-CMO312) built into Cell Ranger. If using official 10x tags, the Feature Reference CSV does not need to list the sequence column; only the CMO tag id values are needed.

csv

id,name,read,pattern,sequence,feature_type
CMO301,CMO301,R2,5P(BC),ATGAGGAATTCCTGC,Multiplexing Capture
CMO302,CMO302,R2,5P(BC),CATGCCAATAGAGCG,Multiplexing Capture
CMO303,CMO303,R2,5P(BC),CCGTCGTCCAAGCAT,Multiplexing Capture
CMO304,CMO304,R2,5P(BC),AACGTTAATCACTCA,Multiplexing Capture
CMO305,CMO305,R2,5P(BC),CGCGATATGGTCGGA,Multiplexing Capture
CMO306,CMO306,R2,5P(BC),AAGATGAGGTCTGTG,Multiplexing Capture
CMO307,CMO307,R2,5P(BC),AAGCTCGTTGGAAGA,Multiplexing Capture
CMO308,CMO308,R2,5P(BC),CGGATTCCACATCAT,Multiplexing Capture
CMO309,CMO309,R2,5P(BC),GTTGATCTATAACAG,Multiplexing Capture
CMO310,CMO310,R2,5P(BC),GCAGGAGGTATCAAT,Multiplexing Capture
CMO311,CMO311,R2,5P(BC),GAATCGTGATTCTTC,Multiplexing Capture
CMO312,CMO312,R2,5P(BC),ACATGGTCAACGCTG,Multiplexing Capture

Download template

At runtime, the cmo_ids used are listed in the [samples] section of the configuration file for demultiplexing (handled by bioinformatics):

csv

[samples]
sample_id,cmo_ids
sample1,CMO301
sample2,CMO303

Using Custom CMO Tags

When using custom CMO tags, provide a complete Feature Reference CSV including id, name, read, pattern, and the actual sequence. Set feature_type to Multiplexing Capture.

Example (modify id / name / pattern / sequence according to your actual experiment):

csv

id,name,read,pattern,sequence,feature_type
HTO1,HTO1,R2,5PNNNNNNNNNN(BC),CTCATTGTAACTCCT,Multiplexing Capture
HTO2,HTO2,R2,5PNNNNNNNNNN(BC),CTGGGCAATTACTCG,Multiplexing Capture

Download template

At runtime, the configuration file specifies the custom cmo-set reference file location and the cmo_ids (handled by bioinformatics):

csv

[gene-expression]
reference,/path/to/transcriptome
cmo-set,/path/to/custom_cmo_ref.csv
create-bam,true
[samples]
sample_id,cmo_ids
sample1,HTO1
sample2,HTO2

If the same sample uses multiple CMO tags (technical replicates), separate the IDs with a pipe |, e.g., CMO301|CMO302. Example:

WARNING

Please inform us in advance about such cases!

csv

[samples]
sample_id,cmo_ids
sample1,CMO301|CMO302
sample2,CMO303|CMO304

Multiplexing Information Checklist (Must Be Provided by Customer)

Information Item	Required
Multiplexing method used (CellPlex CMO / Hashing antibody)	✅ Required
Sample-to-tag mapping table (CMO ID / Hashtag antibody clone)	✅ Required
For custom antibody Hashing: antibody type (TotalSeq-A / B / C) and vendor sequence table	✅ Required (for custom)
Whether any sample uses multiple tags (technical replicates requiring merged output)	Recommended to facilitate runtime configuration

CRISPR Guide Capture

CRISPR Guide Capture detects sgRNAs (protospacers) introduced into individual cells. The Feature Barcode is the sgRNA protospacer sequence itself.

CRISPR products have 8 information columns (2 additional columns beyond the common 6):

Column	Description	Required
`id`	Unique feature ID. Must not contain spaces, slashes, quotes, commas, or other special characters. Must not conflict with transcriptome gene names.	✅ Required
`name`	Feature display name. Can be the same as `id`. Uniqueness not required. Displayed in `matrix/features.tsv.gz`. Same special character restrictions apply.	✅ Required
`read`	`R2`	✅ Required
`pattern`	Fixed sequence indicating the barcode position within the read. Typically 12-20 bp.	✅ Required
`sequence`	The sgRNA protospacer sequence. Must be unique across all features.	✅ Required
`feature_type`	`CRISPR Guide Capture`	✅ Required
`target_gene_id`	Target gene ID, e.g., `ENSG00000113812`	Optional
`target_gene_name`	Target gene name, e.g., `ACTR8`	✅ Required

Filling Guidelines

The sequence column should contain the protospacer sequence of each sgRNA, excluding the scaffold sequence. The length must match the actual design.
If the target_gene_id column is entirely empty, the pipeline will run but perturbation effect analysis will not be performed.
target_gene_id and target_gene_name must match the gene_id and gene_name in the genome GTF file. They are validated during analysis; entries not found in the genome will cause analysis failure.
5' CRISPR requires special attention to sgRNA sequence orientation: The sequencing read direction is 5' to 3'. However, in 5' CRISPR Guide Capture libraries, the sgRNA is captured in the reverse orientation. Therefore, the sequence in the Feature Reference CSV must be the reverse complement of the original design sequence, not the forward sequence. For 5' CRISPR Guide Capture experiments, please confirm whether you are providing the "original sequence" or the "converted sequence."
sgRNA sequences must not be duplicated. If different genes share identical sgRNA sequences by accident, please check and communicate with us in advance to avoid analysis failure or ambiguous counting.
Confirm whether Non-Targeting controls are included. If so, list them separately and clearly label them as Non-Targeting in the id / name fields for downstream analysis.

3' CRISPR Example (fill in according to your actual experiment):

csv

id,name,read,pattern,sequence,feature_type,target_gene_id,target_gene_name
ACTR8-1,ACTR8-1,R2,(BC)GTTTAAGAGCTAAGCTGGAA,GAAGGGCGGCGAGAAGGAGA,CRISPR Guide Capture,ENSG00000113812,ACTR8
ACTR8-2,ACTR8-2,R2,(BC)GTTTAAGAGCTAAGCTGGAA,GAGAACGGAAAGGAGAAGGG,CRISPR Guide Capture,ENSG00000113812,ACTR8
NEG_CTRL-1,NEG_CTRL-1,R2,(BC)GTTTAAGAGCTAAGCTGGAA,GACCGGGGGGGTGCGATGTA,CRISPR Guide Capture,Non-Targeting,Non-Targeting
NEG_CTRL-2,NEG_CTRL-2,R2,(BC)GTTTAAGAGCTAAGCTGGAA,GTGTACTAGTGACGACTATA,CRISPR Guide Capture,Non-Targeting,Non-Targeting

Download template

5' CRISPR Example (fill in according to your actual experiment):

csv

id,name,read,pattern,sequence,feature_type,target_gene_id,target_gene_name
ACTR8-1,ACTR8-1,R2,TTCCAGCTTAGCTCTTAAAC(BC),TCTCCTTCTCGCCGCCCTTC,CRISPR Guide Capture,ENSG00000113812,ACTR8
ACTR8-2,ACTR8-2,R2,TTCCAGCTTAGCTCTTAAAC(BC),CCCTTCTCCTTTCCGTTCTC,CRISPR Guide Capture,ENSG00000113812,ACTR8
NEG_CTRL-1,NEG_CTRL-1,R2,TTCCAGCTTAGCTCTTAAAC(BC),TACATCGCACCCCCCCGGTC,CRISPR Guide Capture,Non-Targeting,Non-Targeting
NEG_CTRL-2,NEG_CTRL-2,R2,TTCCAGCTTAGCTCTTAAAC(BC),TATAGTCGTCACTAGTACAC,CRISPR Guide Capture,Non-Targeting,Non-Targeting

Download template

Pattern Sequence Confirmation Example

This example uses a 5' + CRISPR product based on Addgene: lentiCRISPR v2. The pattern sequence can be confirmed on the Addgene website.

Navigate to the sequence page and locate the sgRNA scaffold sequence:

The basic structure of the gRNA transcript from 5' to 3' is:

5' -- guide sequence (spacer) + sgRNA scaffold -- 3'

guide sequence: The variable region designed to target a specific gene.
sgRNA scaffold: The constant backbone sequence responsible for Cas9 protein binding.

Therefore, the pattern sequence is located at the 5' end of the scaffold. Considering that the length should be moderate (too long may be affected by sequencing errors, too short may not uniquely identify the sequence), we selected the 20 bp GTTTTAGAGCTAGAAATAGC upstream of the scaffold as the pattern. For 5' + CRISPR experiments (reverse capture), the reverse complement is used, resulting in the final pattern GCTATTTCTAGCTCTAAAAC(BC).

Pre-Submission Checklist

[ ] File is UTF-8 / ASCII plain text CSV, no full-width Chinese commas, no BOM header.
[ ] Header columns are strictly id,name,read,pattern,sequence,feature_type in that order.
[ ] id is unique across the entire file, no spaces, commas, or pipes.
[ ] read and pattern match the selected library chemistry (3' TotalSeq-A / B, 5' TotalSeq-C, CRISPR, PTG).
[ ] sequence has been verified for correctness, length matches the expected length of (BC) in the pattern.
[ ] feature_type is correct: Antibody Capture -> Antibody Capture; CRISPR -> CRISPR Guide Capture; Multiplexing (Hashtag) -> Antibody Capture; Multiplexing (CMO) -> Multiplexing Capture.
[ ] If both antibody detection and sample multiplexing are included, both are in the same Feature Reference CSV.
[ ] Confirm the specific product combination (3' / 5', V(D)J or not, CRISPR or not, multiplexing or not) and the sample-to-tag mapping table.

If you are unsure about any of the above, please contact us before submitting to avoid rework due to file format or field errors.

Feature Reference CSV Filling Guide for Antibody Capture, CRISPR Guide Capture, and Cell Multiplexing ​

General Notes for Filling the Reference Table ​

Product Overview ​

Basic Concepts (Common to All Products) ​

Pattern Syntax ​

General Filling Guidelines (Must Check) ​

Antibody Capture ​

TotalSeq-A (Compatible with 3' v2 / v3 kits) ​

TotalSeq-B (Compatible with 3') ​

TotalSeq-C (Compatible with 5' / V(D)J) ​

Cell Multiplexing ​

Antibody Capture Hashing (Most Common, Surface Protein-Based Multiplexing) ​

Multiplexing with Surface Protein Hashtags Only ​

Both Antibody Capture and Hashtag Multiplexing ​

Bioinformatics Notes ​

3' CellPlex with CMO ​

Using 10x Official CMO Tags ​

Using Custom CMO Tags ​

Multiplexing Information Checklist (Must Be Provided by Customer) ​

CRISPR Guide Capture ​

Filling Guidelines ​

Pattern Sequence Confirmation Example ​

Pre-Submission Checklist ​

References ​

Feature Reference CSV Filling Guide for Antibody Capture, CRISPR Guide Capture, and Cell Multiplexing

General Notes for Filling the Reference Table

Product Overview

Basic Concepts (Common to All Products)

Pattern Syntax

General Filling Guidelines (Must Check)

Antibody Capture

TotalSeq-A (Compatible with 3' v2 / v3 kits)

TotalSeq-B (Compatible with 3')

TotalSeq-C (Compatible with 5' / V(D)J)

Cell Multiplexing

Antibody Capture Hashing (Most Common, Surface Protein-Based Multiplexing)

Multiplexing with Surface Protein Hashtags Only

Both Antibody Capture and Hashtag Multiplexing

Bioinformatics Notes

3' CellPlex with CMO

Using 10x Official CMO Tags

Using Custom CMO Tags

Multiplexing Information Checklist (Must Be Provided by Customer)

CRISPR Guide Capture

Filling Guidelines

Pattern Sequence Confirmation Example

Pre-Submission Checklist

References