Genomic Read Summary

segul read summary calculates summary statistics for raw sequence reads in FASTQ format. For assembled contigs, see Genomic Contig Summary.

Input formats

• Gunzip Compressed FASTQ (.fq.gz/.fastq.gz), including BGZF
• Standard FASTQ (.fq/.fastq)

SEGUL detects the compression from the file content. Each file is read once and summarized on all CPU cores, and BGZF files are also decompressed in parallel.

Usage

segul read summary -i raw_reads.fq.gz

or using directory input:

segul read summary -d /raw_read_dir

The output file will be saved as default-read-summary.csv in Read-Summary directory.

To specify the output directory, use the --output or -o argument:

segul read summary -d /raw_read_dir -o read_dir_output

You can also use the --prefix argument to specify the prefix name of the output files:

segul read summary -d /raw_read_dir -o read_dir_output --prefix my_read_summary

Summary modes

SEGUL provides three modes to calculate summary statistics for raw reads:

• Minimal: read count only
• Default: generate essential statistics (see below)
• Complete: generate all the essential statistics plus statistics per read position, distributions, duplication, overrepresented sequences, and adapter content (see below).

Use the --mode option to choose minimal, default, or complete:

segul read summary -d /raw_read_dir --mode complete

Statistics

Essential statistics for raw reads:

• Number of reads
• Number of bases
• Mean read length
• Minimum read length
• Maximum read length
• GC count
• GC content
• AT count
• AT content
• A, C, G, T, N count
• Low-quality base count
• Mean base quality
• Min base quality
• Max base quality
• Median, standard deviation, and N50 of read length
• Bases with quality 20 or higher (Q20) and 30 or higher (Q30)
• Mean quality per read, and reads with mean quality below 20
• Quality score offset (Phred+33 or Phred+64)

Complete mode adds:

• Quality quartiles, 10th and 90th percentiles, and base composition, including N, at each read position (read-pos-summary.zip)
• Read length, mean read quality, and GC content distributions, with the theoretical GC distribution (read-dist-summary.zip)
• Duplication across every read of each file, with the number of copies of each sequence (see below)
• Overrepresented sequences, above 0.1% of all reads, with the adapter they match. Reads longer than 75 bp are compared by their first 50 bp, like FastQC
• Adapter content by position for Illumina Universal, Nextera, Illumina Small RNA, poly-A, and poly-G
• All of the above in read-report.json, which SEGUI opens as an interactive report

Duplication

SEGUL compares the full sequence of every read in the file. FastQC, by comparison, follows only the first 100,000 distinct sequences, so long or sorted files can be misjudged. Counting every distinct sequence of a large file exactly takes a lot of memory, so by default SEGUL keeps a uniform sample of 100,000 distinct sequences and counts their copies across all reads. The duplicate share is estimated from this sample with an error of about 0.3% of the distinct sequences, and the levels chart uses the sample’s exact copy counts. Files with fewer distinct sequences are counted exactly. The sample uses about 30 bytes per sequence, so the default takes about 3 MB per file.

Use --dup-sketch-size to change the sample size. Larger samples are more accurate, and 0 counts every distinct sequence exactly:

segul read summary -d /raw_read_dir --mode complete --dup-sketch-size 1000000

The duplication results are the same with any number of threads. Overrepresented sequences are found among all reads, and their counts can be short by up to 0.025% of reads in files with many distinct sequences, which read-report.json lists as max_error_pct.

Quality score offset

SEGUL detects the quality score offset from the first reads. Use --phred 33 or --phred 64 to set it.

Charts

Add --plot to also write charts to <OUTPUT>/plots. Complete mode adds charts of quality by position (pos-qscore-box_<sample>), read length (length-dist), mean read quality (read-qscore-dist), GC content (gc-dist_<sample>), N content (pos-n-content), adapter content (adapter-content_<sample>), and duplication (duplication_<sample>):

segul read summary -d /raw_read_dir --mode complete --plot

All options

Option	Description
-d, --dir	Directory with FASTQ files
-i, --input	FASTQ files. Allows wildcards
-f, --input-format	Input format: auto (default), fastq, or gzip
--mode	Summary mode: default (default), minimal, or complete
--phred	Quality score offset: auto (default), 33, or 64
--dup-sketch-size	Distinct sequences kept to estimate duplication. Default: 100000. 0 counts all
-o, --output	Output directory. Default: Read-Summary
--prefix	Prefix of the output files
--force	Overwrite an existing output directory
--log	Log file path. Default: segul.log