Notice

1. As "alloca" is used in the source code. It needs more stack memory. Plase check the stack size before running minicom using the command ulimit -s. If it is very small, please change it by ulimit -s unlimited. We are grateful to Prof. Sebastian Deorowicz for reporting the bug.

2. Tomasz Kowalski and Szymon Grabowski reported that minicom fails to correctly decompress several datasets in they work PgRC. We fixed it in the commit 28a736a. The bug is caused when construting contigs. We re-ran some datasets and found that the compression ratio is not affected much (see following table).

Dataset	Compressed file size reported with bug	Latest compressed file size
SRR445718	126064640	126085120
SRR445724	245585920	245585920
SRR445726	215572480	215582720
SRR490961	122552320	122552320
SRR490976	146636800	146575360

minicom

Minicom is a tool for compressing short reads in FASTQ. The minicom program is written in C++11 and works on Linux. It is availble under an open-source license.

Note: Minicom only compresses DNA sequences in the FASTQ file. It does not support to compress the whole FASTQ file.

Download & Install

git clone https://github.com/yuansliu/minicom.git
cd minicom
sh install.sh

In the script install.sh, it downloads the tools bsc and p7zip. Please make sure the two tools can be ran on your machine.

Usage

To compress:

./minicom -r IN.fastq 					
./minicom -r IN.fastq -p 				#order-preserving mode
./minicom -1 IN_1.fastq -2 IN_2.fastq 			#preserving paired-end information		

Minicom creates compressed file IN_comp.minicom, IN_comp_order.minicom and IN_comp_pe.minicom respectively.

Options for compression

-h 		print help message
-t 		number of threads, default: 24 
-r 		reads file 
-1 		reads file of paired-end reads [used for paired-end]
-2 		reads file of paired-end reads [used for paired-end]
-k 		length of k-mer, k <= 31, default: 31
-e 		difference threshold, default: 4
-m 		number of minimizer, default 6
-w 		window size, default L/2-k, where L is the length of reads
-s 		number of indexed substring, default L/17 for L > 80; otherwise L/11
-S 		step size of threshold, default S = e
-E 		maximum threshold, default L/2
-p 		order-preserving mode [only used for compressing single FASTQ file]

To decompress:

./minicom -d IN.minicom [-t number of threads, default 24]

Citation

Yuansheng Liu, Zuguo Yu, Marcel E. Dinger, Jinyan Li; Index suffix-prefix overlaps by (w, k)-minimizer to generate long contigs for reads compression. Bioinformatics, 35(12):2066-2074, 2019.

Contacts

If any bugs during you run our code, please email to yyuanshengliu@gmail.com

---