What's the parameter of UMI-guided assembly

[Angieiscute]

When we are using the following command, we notice the reads with the same UMI did not assemble into 1 read. How can we perform the UMI-guided assembly?
Thanks!

mixcr analyze generic-amplicon-with-umi \
--species hsa \
--tag-pattern "^(R1:*)\^N{1:19}(R2:*)" \
--rna \
--rigid-left-alignment-boundary \
--floating-right-alignment-boundary C \
input_R1.fastq.gz \
input_R2.fastq.gz \
result

[mizraelson]

Where does the UMI located in the read?

[Angieiscute]

the UMI is 14bp and it starts from the 20th bp on R2.

[mizraelson]

Then the command should be:

mixcr analyze generic-amplicon-with-umi \
--species hsa \
--tag-pattern "^(R1:*)\^N{19}(UMI:N{14})(R2:*)" \
--rna \
--rigid-left-alignment-boundary \
--floating-right-alignment-boundary C \
input_R1.fastq.gz \
input_R2.fastq.gz \
result

[Angieiscute]

But with this parameter, the reads with the same UMI will not be assembled into one read. We need to perform UMI-guided assembly which means, all reads are grouped by their UMI.

[mizraelson]

Yes, the command does exactly this. For each UMI group a consensus is created.

[Angieiscute]

We found a fact that maybe the consensus is created, but the count within the group is not corrected into 1 read.

[mizraelson]

Can you show an example? The number of reads stays the same, but you can refer to the number of UMIs. It does not become a single read as it is still important to know that some UMIs have more reads than the others. But the errors are corrected and you can use the UMI count (uniqueMoleculeCount) instead of the read count.

[Angieiscute]

Thanks for your help. uniqueMoleculeCount and uniqueMeculeFraction can be referred as the result after the Consensus assembled?

[mizraelson]

Yes. ’uniqueMoleculeCount’ is the number of unique UMI groups for the clone, which is the number of consensuses (collapsed reads).