Usage

Word length

By default, the word length is 24 nucleotides, this can be changed with the -n flag. The number of nucleotides will be equally divided over all input files. If this is not possible, the remainder is taken from the last input file. For example, if three input files are specified, and -n is set to 23, 7 nucleotides will be taken from the first and second input file, and 9 from the last input file.

If the UMI is present in the header, all nucleotides from the UMI will be used, and the remainder will be divided between the input files as described.

UMIs in separate file

As this tool can handle an arbitrary number of input files, simply specify the file with the UMIs as an input file. Note that with a separate UMI file, there will be no preference to include all UMI nucleotides.

humid forward.fastq.gz reverse.fastq.gz umi.fastq.gz

UMIs in the header with underscore

If UMIs are include in the header, the file with the UMI in the header must be the first specified input file, and the UMI must be before any spaces.

Below are some examples of headers with the UMI AGTA after an underscore, which will all be recognised.

@A31886:289:T5D5W10Y2:2:12686:4678:1110_AGTA
@1_AGTA
@1_AGTA with spaces after

This format of specifying UMIs in the read header is used by UMI-Tools.

UMIs in the header with colon

If UMIs are include in the header, the file with the UMI in the header must be the first specified input file, and the UMI must be before any spaces.

Below are some examples of headers with the UMI AGTA that will be recognised.

@A31886:289:T5D5W10Y2:2:12686:4678:1110:AGTA
@A31886:289:T5D5W10Y2:2:12686:4678:1110:AGTA with spaces after
@1:::::::AGTA

This format of specifying UMIs in the read header is used by BCL Convert and fastp.

Other UMI formats

If your data does not confirm to any of the supported schemas described above, we recommend fastp to process the UMI to one of the supported formats.

Deduplication without UMI

If a project was sequenced without UMIs, you can still remove duplicates using this tool, since it will automatically take the requested number of nucleotides from one (single end) or two (paired end) FastQ files. Note that without random UMIs to distinguish identical but independent molecules, the number of duplicates will most likely be an overestimation, similar to using picard MarkDuplicates.