Assemble clones¶

The assemble command builds a set of clones using alignments obtained with align command in order to extract specific gene regions (e.g. CDR3). The syntax of assemble is the following:

mixcr assemble [options] alignments.vdjca output.clns

The following flowchart shows the pipeline of assemble:

This pipeline consists of the following steps:

The assembler sequentially processes records (aligned reads) from input .vdjca file produced by align. On the first step, assembler tries to extract gene feature sequences from aligned reads (called clonal sequence) specified by assemblingFeatures parameter (CDR3 by default); the clonotypes are assembled with respect to clonal sequence. If aligned read does not contain clonal sequence (e.g. CDR3 region), it will be dropped.
If clonal sequence contains at least one nucleotide with low quality (less than badQualityThreshold parameter value), then this record will be deferred for further processing by mapping procedure. If percent of low quality nucleotides in deferred record is greater than maxBadPointsPercent parameter value, then this record will be finally dropped. Records with clonal sequence containing only good quality nucleotides are used to build core clonotypes by grouping records by equality of clonal sequences (e.g. CDR3). Each core clonotype has two main properties: clonal sequence and count — a number of records aggregated by this clonotype.
After the core clonotypes are built, MiXCR runs mapping procedure that processes records deferred on the previous step. Mapping is aimed on rescuing of quantitative information from low quality reads. For this, each deferred record is mapped onto already assembled clonotypes: if there is a fuzzy match, then this record will be aggregated by the corresponding clonotype; in case of several matched clonotypes, a single one will be randomly chosen with weights equal to clonotype counts. If no matches found, the record will be finally dropped.
After clonotypes are assembled by initial assembler and mapper, MiXCR proceeds to clustering. The clustering algorithm tries to find fuzzy matches between clonotypes and organize matched clonotypes in hierarchical tree (cluster), where each child layer is highly similar to its parent but has significantly smaller count. Thus, clonotypes with small counts will be attached to highly similar “parent” clonotypes with significantly greater count. The typical cluster looks as follows:

After all clusters are built, only their heads are considered as final clones. The maximal depths of cluster, fuzzy matching criteria, relative counts of parent/childs and other parameters can be customized using clusteringStrategy parameters described below.
The final step is to align clonal sequences to reference V,D,J and C genes. Since the assemblingFeatures are different from those used in align, it is necessary to rebuild alignments for clonal sequences. This alignments are built by more accurate aligner (since all hits are known in advance); thus, better alignments will be built for each clonal sequence.
The result is written to the binary output file (.clns) with a comprehensive information about clones.

Command line parameters¶

The command line options of assemble are the following:

Option	Default value	Description
`-h`, `--help`		Print help message.
`-r`, `--report`		Report file name. If this option is not specified, no report file be produced.
`-t`, `--threads`	number of available CPU cores	Number of processing threads.
`-Oparameter=value`		Overrides default value of assembler `parameter` (see next subsection).

All parameters are optional.

Assembler parameters¶

MiXCR uses a wide range of parameters that controls assembler behaviour. There are some global parameters and parameters organized in groups for each stage of assembling: cloneClusteringParameters and cloneFactoryParameters. Each group of parameters may contain further subgroups of parameters etc. In order to override some parameter value one can use -O followed by fully qualified parameter name and parameter value (e.g. -Ogroup1.group2.parameter=value).

One of the key MiXCR features is ability to assemble clonotypes by sequence of custom gene region (e.g. FR3+CDR3); target clonal sequence can even be disjoint. This region can be specified by assemblingFeatures parameter, as in the following example:

mixcr assemble -OassemblingFeatures="[V5UTR+L1+L2+FR1,FR3+CDR3]" alignments.vdjca output.clns

(note: assemblingFeatures must cover CDR3).

Other global parameters are:

Parameter	Default value	Description
`badQualityThreshold`	`20`	Minimal value of sequencing quality score: nucleotides with lower quality are considered as “bad”. If sequence contains at least one “bad” nucleotide, it will be deferred at initial assembling stage, for further processing by mapper.
`maxBadPointsPercent`	`0.7`	Maximal allowed percent of “bad” points in sequence: if sequence contains more than `maxBadPointsPercent` “bad” nucleotides, it will be dropped.
`addReadsCountOnClustering`	`false`	Aggregate cluster counts when assembling final clones: if `addReadsCountOnClustering` is `true`, then all children clone counts will be added to the head clone; thus head clone count will be a total of its initial count and counts of all its children.

One can override these parameters in the following way:

mixcr assemble -ObadQualityThreshold=10 alignments.vdjca output.clns

In order to prevent mapping of low quality reads (filter them off) one can set maxBadPointsPercent to zero:

mixcr assemble -OmaxBadPointsPercent=0 alignments.vdjca output.clns

Clustering strategy¶

Parameters that control clustering procedure are placed in cloneClusteringParameters parameters group:

Parameter	Default value	Description
`searchDepth`	`2`	Maximum number of cluster layers (not including head).
`allowedMutationsInNRegions`	`1`	Maximum allowed number of mutations in N regions (non-template nucleotides in VD, DJ or VJ junctions): if two fuzzy matched clonal sequences will contain more than `allowedMutationsInNRegions` mismatches in N-regions, they will not be clustered together (one cannot be a direct child of another).
`searchParameters`	`twoMismatchesOrIndels`	Parameters that control fuzzy match criteria between clones in adjacent layers. Available predefined values: `oneMismatch`, `oneIndel`, `oneMismatchOrIndel`, `twoMismatches`, `twoIndels`, `twoMismatchesOrIndels`, ... , `fourMismatchesOrIndels`. By default, `twoMismatchesOrIndels` allows two mismatches or indels (not more than two errors of both types) between two adjacent clones (parent and direct child).
`clusteringFilter` `.specificMutationProbability`	`1E-3`	Probability of a single nucleotide mutation in clonal sequence which has non-hypermutation origin (i.e. PCR or sequencing error). This parameter controls relative counts between two clones in adjacent layers: a smaller clone can be attached to a larger one if its count smaller than count of parent multiplied by `(clonalSequenceLength * specificMutationProbability)` `^ numberOfMutations`.

One can override these parameters in the following way:

mixcr assemble -OcloneClusteringParameters.searchParameters=oneMismatchOrIndel alignments.vdjca output.clns

In order to turn off clustering one should use the following parameters:

mixcr assemble -OcloneClusteringParameters=null alignments.vdjca output.clns

Clone factory parameters¶

Parameters which control final alignment of clonal sequences are placed in cloneFactoryParameters group. These parameters includes separate groups for V, D, J and C aligners: vParameters, dParameters, jParameters and cParameters. The D aligner is the same as used in align and thus all its parameters and their default values are the same as described for D aligner in align. One can override these parameters in the following way:

mixcr assemble -OcloneFactoryParameters.dParameters.absoluteMinScore=10 alignments.vdjca output.clns

mixcr assemble -OcloneFactoryParameters.dParameters.scoring.gapOpenPenalty=-10 alignments.vdjca output.clns

The aligners used to build alignments with V, J and C genes are different from those used by align.

V, J and C aligner parameters¶

The following table lists parameters of V, J and C aligners:

Parameter	Default V value	Default J value	Default C value	Description
`featureToAlign`	`VTranscript`	`JRegion`	`CExon1`	Gene region used to build alignments.
`relativeMinScore`	`0.8`	`0.8`	`0.8`	Relative minimal score of hit: hits with score less than `relativeMinScore * maxScore` (`maxScore` is score of best hit) will be dropped.

One can override these parameters in the following way

mixcr assemble -OcloneFactoryParameters.jParameters.featureToAlign=JRegion(-6,0) alignments.vdjca output.clns

The scoring parameters are placed in group alignmentParameters.scoring:

Parameter Default value (same for V, J, C) Description

subsMatrix

simple(match = 5,: mismatch = -9)

Substitution matrix. Available types:

simple — a matrix with diagonal elements equal to match and other elements equal to mismatch

raw — a complete set of 16 matrix elements should be specified; for example: raw(5,-9,-9,-9,-9,5,-9,-9,-9,-9,5,-9,-9,-9,-9,5) (equivalent to the default value)

gapPenalty -12 Penalty for gap.

One can override these parameters in the following way

mixcr assemble -OcloneFactoryParameters.vParameters.alignmentParameters.scoring.gapPenalty=-5 \
               alignments.vdjca output.clns