The Single Cell Antibody Annotator operation is an alternate way to annotate and analyze the variable regions of standard IgG-like molecules. It combines the sequences analyzed into fewer representative sequences and can pull out the dominant chain(s) within your dataset.
The Single Cell Analysis operation is particularly suited for analysing Barcoded Sequences from both "single clones in wells" and "single cells in droplets" experiments. It is also suitable for analyzing datasets from NGS technologies that incorporate UMIs. To learn more about Barcodes and UMIs, see this article: Understanding Single Cell technologies: Barcodes and UMIs. The video below provides an overview of these concepts:
For Barcode and/or UMI analysis, you will first need to run the Collapse UMI Duplicates and Separate Barcodes tool. For more discussion about possible workflows, see the Single Cell Analysis workflows article.
Jump to:
How do I run Single Cell Antibody Annotator?
Select one or more nucleotide sequences or lists in your folder and select Annotation > Single Cell Antibody Anotator:
To run the Single Cell Antibody Annotator operation, select options relevant to your sequence data in the dialog that appears and click Run.
Each selected input sequence list will be analysed independently. The pipeline works by:
- Trimming sequences and then applying filtering steps to reduce total data size (configurable)
- De-novo assembling reads together (optional)
- Identifying and annotating Ig-like Antibody regions (FR/CDR)
- Collapsing sequences with similar V(D)J regions into a single dominant sequence. Heavy and light chains are compared separately.
When run on barcoded sequences in combination with the Collapse UMI Duplicates and Separate Barcodes operation, it can identify heavy-light chains pairs and determine dominant chains within each cell/barcode. For an example of a workflow for analysing barcoded data, please refer to the UMI/Barcode and Single Cell Analysis tutorials.
Single Clone Antibody Annotator options
Main Options
-
The Reference Database can be a database of antibody germline sequences or of template (variable region) sequences. The reference database is used to help identify the correct FR and CDR regions in the new sequences being analyzed. See Understanding Reference Databases.
- Your Biologics account will include a Human, Mouse and Alpaca Ig germline database. If you would like to access to other germlines, please either contact us or see How to make a Custom Reference Database.
Sequence region of interest is between:
- To define what a "fully annotated" sequence is, you can select the values from the dropdown menu. The default values between FR1 and FR4 means that a sequence is considered to be fully annotated if it consists of all of the regions: FR1, CDR2, FR2, CDR2, FR3, CDR3, and FR4.
- This option will collapse "identical" sequences up to a percentage threshold of identity. It is recommended to use an identity percentage low enough to capture sequencing errors but high enough to preserve true variation. 97% is a reasonable default.
- Note that the collapsed sequence includes both the "Sequence region of interest is between" option above, as well as the "Retain upstream/downstream of fully annotated region" bp sequence if selected below.
Retain upstream and downstream of fully annotated region:
- These options retain the specified number of nucleotides upstream/downstream of the Sequence region of interest (explained above) when trimming the ends of contigs. This is in order to identify duplicates when ignoring incorrect contig ends. If a contig is not long enough to cover the specified range, it will be excluded from the next step in the pipeline.
Associate significant dominant heavy and light pair
-
This option is recommended for barcoded data or sequencing data from separate sequence lists that represent identical samples. If heavy and light chains are present under the same barcode/sequence list, a Chain Combinations table will be generated. This table will enumerate the possible heavy/light chain pairings found under the same barcode/sequence list.
- Note that this only pairs the 4 most prevalent heavy and light chains that meet the significance threshold (see the filtering section).
Keep unmerged reads
- This option specifies that paired reads which failed to merge should be used in the next step of the pipeline. In use cases where sequence reads are expected to overlap, discarding unmerged reads is recommended in order to improve assembly accuracy.
De novo assembly
- Select the De novo assembly required option to perform de novo assembly if reads are fully annotated region fragments. Otherwise reads are grouped via the fully annotated region annotation.
Analysis Options
Antibody numbering:
- The default scheme is IMGT CDR definitions and numbering. We support IMGT, Kabat, Chothia, Martin and AHo schemes. To learn more, see Numbering Schemes. You can also turn on and off annotating the numbers on your sequences here.
Annotate variants from reference database
- To annotate and see the differences between your input sequences and reference sequences, select the Annotate germline differences option. With the selection of this option, the nucleotide and amino acid differences will be annotated on your input sequences.
Calculate protein statistics
- This will calculate the Molecular Weight (kDa), the Isoelectric point, the charge at pH 7 and the Extinction Coefficient across the VDJ or VJ region - or both. If a full V(D)J Region can not be found these values will not be calculated.
Find liabilities and assets:
- To search and score amino acid or nucleotide motifs associated with deleterious post-translational modifications or any type of reduced antibody function or desirable motifs, select this option. Biologics has a default set of sequence liability checks, these include: cleavage, deamidation, glycosylation, hydrolysis, isomerization and oxidation. To learn how to specify your own liabilities, see this article: How to customize antibody sequence liabilities and assets
Additional features:
- This option allows you to specify custom features (such as fusions, constant regions, or Signal Peptides) that will be annotated on each sequence. To learn more about annotation of additional features, please see Using Feature Databases to identify Constant Regions and Fusion Proteins.
Filtering Options
Only keep sequences longer than:
- All sequences that are shorter than the threshold defined will be discarded. This is useful to remove likely low quality sequences. You may wish to set this parameter lower if your sequences have adapters, UMIs and/or barcodes which have been trimmed in the Collapse UMI Duplicates and Separate Barcodes operation.
Only use longest:
- This option lets you specify the number of the reads that will be used from each list or barcode (after sorting by length), with any additional reads discarded. This helps improve performance on large data sets where excessive data significantly slows down analysis. 500 reads is generally sufficient.
Only keep regions with at least:
- This input field lets you discard sequences where the number of reads is equal to or less than the number of reads for the most prevalent chain within the barcode/dataset, specified as a percentage. This setting allows you to permanently discard regions that do not have enough supporting data for further analysis.
Significant regions
Significant region thresholds allow you to filter out infrequent regions to retain statistically significant regions for downstream analysis. Regions deemed insignificant are retained but annotated as not significant.
- The Significant regions have at least (percentage read count of the cell/barcode) input field lets you flag sequences with low numbers of reads relative to the total reads in the cell, specified as a percentage.
- The Significant regions have at least (reads) input field lets you flag sequences with low numbers of reads as not significant. This is useful for filtering out sequences that were only defined due to reads of very low frequency or due to sequencing errors.
- The Significant regions have at least (percentage read count of the dominant same chain region) input field lets you flag sequences where the number of reads is equal to or less than the number in the dominant same chain region (the one with the most reads), specified as a percentage. This setting allows you to filter out regions that do not have enough supporting data for further analysis.
Clustering Options
Clustering provides a way to specify clonotype parameters and to group your sequences based on genes or regions of interest. To learn more about clustering and how it can help with interpreting your dataset, see Understanding "Clusters".
Several default clusters will already be listed, and further clusters can be added using the blue "Plus" sign as seen above. Any clusters that are not applicable for a particular analyses, such as Light CDR3 in a Heavy chain data set, will be omitted.
It is possible to create custom clusters with up to six regions or genes (FR1, CDR3, Heavy D gene etc.), allow mismatches across a region, and/or cluster based on amino acid similarity. To learn more about configuring advanced clustering options, please refer to Clustering Options.
Cluster Filters
In general, most NGS datasets are relatively large and contain low quality sequences and noise. In order to improve the meaningfulness of clusters in your results, select the following options:
-
Only cluster results with asset and liability score of at least - This option can help to speed up the run-time for large datasets. Sequences will be clustered based on whether they meet the score specified. For example, if you specify a score of -1000, only sequences that have a liability and asset score of -1000 or more will be included in the clusters.
Advanced Options
Document name scheme:
- This dropdown lets you optionally select a Name Scheme that can read the collapsed single clone names to extract information of interest. Single Cell Antibody Analysis will use this information to pair Heavy/Light chains (if the Name Scheme has chain values) and output all Name Scheme fields as columns in the results table. For more information about Name Schemes, see Using Name Schemes
Genetic code:
- The Genetic code dropdown allows you to select the genetic code to use for translating nucleotide sequences. The codes are obtained directly from NCBI. One additional Genetic code, "Amber readthrough" allows the Amber Stop Codon not to be treated as a stop codon during translation.
Record equal gene matches as:
- Each gene with partial frequency - This will assign the query sequence to all matching genes with partial frequency. Based on the example of a query sequence matching two genes: IGHD1-26 and IGHD2-15, the query sequence will add 0.5 to the total count for both IGHD1-26 and IGHD2-15.
- Group of genes - This will create a separate entry in the list of genes that represents this combination of genes. Based on the example of a query sequence matching two genes: IGHD1-26 and IGHD2-15, the query sequence will contribute 0 towards the total for each of IGHD2-15 and IGHD1-26, and instead add 1 to the total for a gene called "IGHD1-26/IGHD2-15".
- Unknown - This will treat this as an unknown gene. Based on the example of a query sequence matching two genes: IGHD1-26 and IGHD2-15, the query sequence will add nothing to the totals of IGHD1-26 and IGHD2-15.
Pseudogenes and ORF genes
- The human and mouse immunoglobulin databases may contain a number of pseudogenes and ORF genes along with fully functional genes. You can choose to Include pseudogenes from database and/or the ORF genes from database in the analysis as mutations in these genes might be corrected prior to expression. If these are not included, then each sequence will be classified according to the most closely related functional gene in the database.
Trim primers:
- The Trim primers option allows you to select a primer database from the dropdown for trimming any of the primers in the database from your sequences, if they are present.
Always annotate entire regions (except CDR3)
- If a CDR or FR annotation would be truncated due to mismatches, this option instead forces it to end at the boundary of the respective CDR/FR region. That region will be complete and not truncated.
Heavy & Light FR adjustments:
- This option is explained in more detail here: Adjusting CDR definitions