How to troubleshoot barcode collision and not enough hamming distance errors when demultiplexing runs with BCL Convert and bcl2fastq2

What is barcode collision?

An index/barcode collision error occurs when the index sequences in the sample sheet have insufficient Hamming distance to support the selected mismatch settings — meaning the sequences are too similar to be reliably distinguished. When this condition is detected, a sequence read from the instrument cannot be uniquely assigned to a single sample index.

Barcode Collision Example 1: The two index sequences are distinct by 1 base pair. A single mismatch would result in barcode collision.

Barcode Collison example 2: The two index sequences are distinct by 2 base pairs. One mismatch would also result in barcode collision.

What is hamming distance?

Hamming distance measures the minimum number of substitutions required to change one sequence into the other, or the minimum number of errors that could have transformed one sequence into the other.

How to troubleshoot barcode collision error?

If there is a barcode collision/hamming distance error, the number of barcode mismatch will need to be decreased. The default setting is 1 barcode mismatch for bcl2fastq and BCL Convert, but this can be decreased to 0 (no mismatches allowed) or increased to a maximum of 2.

-For BCL Convert:

During Run Preparation (either in BaseSpace or in the Control Software Run Planning), select 0 in the dropdown menu (if no selection is made, the software will consider the default option of 1):

This will be shown in the sample sheet as:

-For bcl2fastq2:

Add the following to the command line:

--barcode-mismatch 0

Note: Barcode mismatches = # of differences allowed in the index sequence when assigning reads to samples.

An important distinction: This is not the number of bases different between indexes in the sample sheet

For any feedback or questions regarding this article (Illumina Knowledge Article #2326), contact Illumina Technical Support [e-mail].