A Demultixper for Combinatorially Multiplexed Next Generation Reads
Author: Daniel E. Lefever
Mr. Demuxy is a fairly simple, self-contained Python program that makes demultiplexing variable length, combinatorially barcoded NGS reads as painless as possible. See below for more details. This program was written with the combinatorial barcoding strategy developed by Travis Glenn in mind [Adapterama Ref], however, it would work on any type of data multiplexed with combinatorial barcodes. In short, this approach uses two pairs of barcodes denoted as the internal and external barcodes (also referred to as tags and indexes). The idea is to pool many different projects onto one MiSeq/HiSeq run cheaply by having external Illumina barcodes to differentiate the pooled projects and internal ‘homebrewed’ variable length barcodes to distinguish samples within each project. The internal barcoding strategy is based on the layout of a 96-well plate: the forward tag corresponds to row letter (A-H) and the reverse tag corresponds to column number (1-12). This program uses both tags to come up with the sample ID.
First Download Mr. Demuxy and unarchive the folder. To unarchive the folder, try clicking on it. If that doesn’t work, open terminal and change directory to wherever it downloaded (probably /Downloads) and try this:
$ tar -xvzf Mr_Demuxy-x.y.z-dist.tar.gz
NOTE: The ‘x.y.z’ above denotes version number. Change that to whatever version is downloaded. Also, relevant xkcd
Then open up terminal and change directory to wherever the unarchived folder lives. Then, install it to your local site-packages directory like so:
$ python setup.py install --user
Restart terminal. You should now be able to run the program from anywhere in your directory.
Can I use pip install?
Yes, but for whatever reason, pip install does not execute some extra stuff I included in the setup script. The extra stuff basically adds a line to your .bash_profile which puts the local script directory in the path. This is so you can call the scripts from anywhere. Also, it will remove some stuff leftover from previous Mr. Demuxy versions. In order to do this manually, you’ll need to add this to your .bash_profile:
Mr. Demuxy contains two main scripts; merged_demuxer.py and pe_demuxer.py for merged or paired-end reads, respectively. Basic usage for each script are outlined below.
Demultiplex Merged Reads (merged_demuxer.py)
usage: $ merged_demuxer.py -f forward_bcs.txt -r reverse_bcs.txt -i example_merged.fastq -o demultiplexed_seqs.fasta This program demultiplexes merged combinatorially barcoded reads and can output the demultiplexed reads a number of different ways. required arguments: -f forward bcs file -r reverse bcs file -i input merged fastq file optional arguments: -o name of output file or folder (default: merged_demuxer_out) --forward_primer_len base number of forward primer/adapter (default: 0) --reverse_primer_len base number of reverse primer/adapter (default: 0) --no_rev_comp_reverse_bcs prevents rev bcs from reverse complementing -h, --help show this help message and exit --keep_original_headers pass this arg to keep original headers in demultiplexed fasta/fastq file --write_qual write a qual file containing quality scores --individual_fastq demultiplex by writing a fastq file for each sample ID --min_seq_length minimum length of seq to be retained (default: 20) --file_extension format of the output file (default: fasta)
Demultiplex Paired-End Reads (pe_demuxer.py)
usage: $ pe_demuxer.py -r1 r1_seqs.fastq -r2 r2_seqs.fastq -r1_bc r1_bc_file.txt -r2_bc -r2_bc_file.txt This program splits combinatorially multiplexed paired-end seqs and outputs a folder for the R1 reads and another for the R2 reads. The two folders contain one fastq file per sample id. required arguments: -r1_bc BC file for R1 reads -r2_bc BC file for R2 reads -r1 R1 fastq file -r2 R2 fastq file optional arguments: -h, --help show this help message and exit -o name of output file or dir --batch_length amount of seqs to be put in a batch only used when check_headers is passed (Default: 1500000) --forward_primer_len base number of primer/adapter on R1 seq (default: 0) --reverse_primer_len base number of primer/adapter on R2 seq (default: 0) --keep_original_headers pass this arg to keep original headers in demultiplexed fasta/fastq file --min_seq_length minimum length of seq to be retained (default: 20) --check_headers ensures the R1 and R2 read match before demultiplexing WARNING: this is super slow
Usage Examples: merged_demuxer.py
Demultiplex merged fastq reads with output being in fasta format:
$ merged_demuxer.py -i example_merged.fastq -f forward_bcs.txt -r reverse_bcs.txt -o example_demult.fasta
Do the same as above and get a seperate qual file:
$ merged_demuxer.py -i example_merged.fastq -f forward_bcs.txt -r reverse_bcs.txt -o example_demult.fasta --write_qual
Demultiplex and keep the original fastq headers:
$ merged_demuxer.py -i example_merged.fastq -f forward_bcs.txt -r reverse_bcs.txt -o example_demult.fasta --keep_original_headers
Demultiplex merged reads and remove the 17 bp forward primer and 22 bp reverse primer:
$ merged_demuxer.py -i example_merged.fastq -f forward_bcs.txt -r reverse_bcs.txt -o example_demult.fasta --forward_primer_len 17 --reverse_primer_len 22
Do the same as above except have the output be in fastq format:
$ merged_demuxer.py -i example_merged.fastq -f forward_bcs.txt -r reverse_bcs.txt -o example_demult.fastq --forward_primer_len 17 --reverse_primer_len 22 --file_extension fastq
Demultiplex reads and split into one fastq file per sample ID:
$ merged_demuxer.py -i example_merged.fastq -f forward_bcs.txt -r reverse_bcs.txt -o example_demult_ind_fastq --individual_fastq
Get individual fastq files, chop off the 17 bp forward primer and 22 bp reverse primer, and keep reads that are only longer than 100 bp:
$ merged_demuxer.py -i example_merged.fastq -f forward_bcs.txt -r reverse_bcs.txt -o example_demult_ind_fastq --individual_fastq --forward_primer_len 17 --reverse_primer_len 22 --individual_fastq --min_seq_length 100
Usage Examples: pe_demuxer.py
Demultiplex paired-end fastq files and write individual fastqs for each sample in their respective directories:
$ pe_demuxer.py -r1 r1_seqs.fastq -r2 r2_seqs.fastq -r1_bc r1_bc_file.txt -r2_bc -r2_bc_file.txt
Do the same as above and remove a 10 bp primer on R1 reads and a 16 bp primer on R2 reads:
$ pe_demuxer.py -r1 r1_seqs.fastq -r2 r2_seqs.fastq -r1_bc r1_bc_file.txt -r2_bc -r2_bc_file.txt --forward_primer_len 10 --reverse_primer_len 16
Demultiplex paired-end reads in which the read order may be different and/or the files contain un-paired reads: WARNING: using this option is extremely slow and not recommended. Illumina sequencers should produce files with reads in the same order and discard un-paired reads. I didn’t realize that until after added this feature so I decided to include it anyways.
$ pe_demuxer.py -r1 r1_seqs.fastq -r2 r2_seqs.fastq -r1_bc r1_bc_file.txt -r2_bc -r2_bc_file.txt --check_headers
Input Files Information:
This section explains more about the input files.
Forward or R1 barcode file The forward barcode file, as you might expect, is the file containing the forward barcodes and letter (or whatever the barcode corresponds to). You will probably need to make your own forward barcode file, and the formatting is important. The easiest way to make one, is to open a new sheet in excel and copy/paste the barcode and letter pairs with the first column being the letter and the second containing the barcode. It should look like this:
A GGTAC B CAACAC C ATCGGTT D TCGGTCAA
Note: Both barcode files do not have any header information. I’m not sure if it will screw things up, but it’s safest to not include it. Once you’re done inputing the forward barcodes, save the excel sheet as a ‘tab delineated text’ file. Be sure to name it something so you know what it is.
Reverse or R2 barcode file The reverse barcode file is just like the forward barcode file, except for the reverse barcodes. It should look like this:
1 AGGAA 2 GAGTGG 3 CCACGTC 4 TTCTCAGC
Note: This program reverse complements (RC) the reverse barcodes when demultiplexing merged reads instead of RCing the whole sequence. This makes it go faster, but if your reverse barcodes do not need to be RC’d, pass this:
Output File Information:
The output file/files contain reads that have been demultiplexed, and optionally trimmed of primers. This program automatically removes the barcode from each read and will remove the primers if length is specified by the –forward_primer_len and –reverse_primer_len arguments. The primers are not actually matched and are instead removed by trimming a set number of bases after the barcode. This is much quicker than actual matching.
For example, if the forward primer was 17 bp long, and the reverse 15, pass:
--forward_primer_len 17 --reverse_primer_len 15
If these options are not passed, it defaults to 0.
The merged_demuxer.py script can also create individual fastq files split by each sample. To do this, simply pass the –individual_fastq argument. The -o argument will become the name of the output directory with one fastq per sample. The dir structure is as follows:
output_dir |---<sample_id_1>.fastq |---<sample_id_2>.fastq . . |---<sample_id_n>.fastq
pe_demuxer.py will only output individual fastq files. The R1 and R2 files are separated into two folders each containing individual fastq files split by sample. This is the structure:
output_dir |---R1 | |---<sample_id_1>_R1.fastq | |---<sample_id_2>_R1.fastq | . | . | |---<sample_id_n>_R1.fastq | | |---R2 |---<sample_id_1>_R1.fastq |---<sample_id_2>_R1.fastq . . |---<sample_id_n>_R1.fastq
The merged_demuxer.py script offers some control over how the headers are formatted. If there’s some specific way that would be helpful, please let me know and I’ll include it in future releases. Anyways, here are the current options for the header format:
HEADER OPTIONS: -Demultiplexed ID only- --FASTA Format-- >A6_1 AATATT....CACGCAGA --FASTQ Format-- @A6_1 AATATT....CACGCAGA + GGGGGG...@GGFGGFC< -Original header with demultiplexed ID- --FASTA Format-- >M02849:...:1886 A6_1 AATATT....CACGCAGA --FASTQ Format-- @M02849:...:1886 A6_1 AATATT....CACGCAGA + GGGGGG...@GGFGGFC<
This program was written and predominantly tested using a macbook pro 2011 with 8 GB memory and a 2.4 GHz Intel Core i7 processor running OS X El Capitan using python 2.7.10. With this setup, both merged (written to single or individual files) and paired-end demultiplexing of ~7,000,000 reads took around 3-4 minutes.
the fastq parser was taken from biopython
Cock, P. J., Antao, T., Chang, J. T., Chapman, B. A., Cox, C. J., Dalke, A., … de Hoon, M. J. (2009). Bio$ python: freely available $ python tools for computational molecular biology and bioinformatics. Bioinformatics, 25(11), 1422-1423. doi:10.1093/bioinformatics/btp163
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.