OPTIONS
-k, --kmers-length
size of kmers [default:21]

-w, --window-size
size of window guaranteed to contain at least one minimizing kmer [default:40]

-d, --density
on average 1/2^d kmers are indexed [default:3]

-t, --threads
number of threads [default:1]

-a, --dropout
QD does not try to deconvolve clouds smaller than this value [default:0]

```

### Input

QuickDeconvolution takes as input `-i` a fasta or a fastq file containing barcoded reads with the tag `BX:Z` designating a barcode (this is the default output of [longranger basic](https://support.10xgenomics.com/genome-exome/software/pipelines/latest/advanced/other-pipelines)). For example
```
@read_456 cov:23.45 BX:Z:AAAACTGTAT
```
If the reads are paired, provide QuickDeconvolution with an interleaved file where the two ends of the pairs have the same name, it will recognize it. To interleave two files, you can use this command line:
```
paste -d '\n' <(awk '{if (NR%4==1){printf"\n";printf $0;} else{printf "((()))"$0;}}' reads_foward.fq) <(awk '{if (NR%4==1){printf"\n";printf $0;}else{ printf "((()))"$0;}}' reads_reverse.fq) | sed 's/((()))/\n/g' > sequencing_reads_interleaved.fastq
```

### Output

QuickDeconvolution outputs the fasta/q file given as input, with an additional tag (-0, -1, -2...) at the end of the line, so that the deconvolved reads look like
```
@read_456 cov:23.45 BX:Z:AAAACTGTAT-1
```
Within each barcode, reads having the same tag come from the same fragment. WARNING: the -0 tag is a special tag, indicating reads that could not be deconvolved by the program. If a tag is already present, QuickDeconvolution will nonetheless append a new tag at the end of the barcode:
```
@read_456 cov:23.45 BX:Z:AAAACTGTAT-1-3
```

### Options

Option -a is the dropout option: the program disregard all clouds containing fewer reads than this value. You may want to use the option if you know you'll need clouds of a certain size for your downstream analyses, in which case it might be a waste of time to deconvolve the smallest clouds.

Option -t is the number of threads to launch simultaneously on the program. Wall-clock time decreases and RAM usage increases with the number of threads.

Options k, w and d are parameters of the alignment within QuickDeconvolution. The deconvolution should not be very sensitive to these values.
k is the length of the k-mers. Avoid decreasing k below 15.
d is to monitor the density of sparse k-mers. On average 1/2^d k-mers will be sparse.
While choosing sparse k-mers, the program is ensured to choose at least 1 k-mer in a window of size w.
Decreasing w and/or d may in some cases increase precision at the expense of run-time. Keep w within the range [10,50] and d within range [1,5].

## License

QuickDeconvolution is distributed under the license GPL3

## Citation

QuickDeconvolution is published in *Bioinformatics advances*. You can cite using:
Faure, Roland, and Dominique Lavenier. “QuickDeconvolution: Fast and Scalable Deconvolution of Linked-Read Sequencing Data.” *Bioinformatics Advances*, September 26, 2022, vbac068. _https://doi.org/10.1093/bioadv/vbac068_.