Rasmus Amund Henriksen, Lei Zhao, Thorfinn Sand Korneliussen
Contact: rasmus.henriksen@sund.ku.dk
This article was published in Oxford Academics as an application note the 20th of January 2023.
NGSNGS is a new program, therefore we are very interested in feedback to solve potential problems, as well as ideas for improvements or additions to specific and relevant features.
Additions, Improvements, Alterations
NGSNGS requires HTSlib, a library used for handling high-throughput sequencing data.
git clone https://github.com/samtools/htslib.git
git clone https://github.com/RAHenriksen/NGSNGS.git
cd htslib
git submodule update --init --recursive
make
cd ../NGSNGS; make HTSSRC=../htslib
To install NGSNGS do:
git clone https://github.com/RAHenriksen/NGSNGS.git
cd NGSNGS; make
make amplicon HTSSRC=../htslib
NOTE: Newer version of htslib which includes bam_set1 is required
The following features has been added since its publication in January 2023. See helppage for usage Help page
- Fixed quality score (-qs): creating one fixed quality score with identical sequencing error for each nucleotide.
- Circular simulation mode (-circ): enable the possibility of simulating sequencing reads crossing the reference genome end coordinate - to simulate bacterial or mitochondria amongst others
- Stochastic mutation rate (-mr) : adding stochastic variation to the input reference genome (-i)
- Fixed number of stochastic SNP's (-v): adds a fixed number of randomg mutations ot the input reference genome (-i)
- Simulate reads solely for regions of interest across the genome (-incl): the coordinates from which the reads are sampled needs to be in a bed file structure. With the possibility of adding a number of nucleotides in both the start and end of the region by providing the flanking option -fl. This can be used for CAPTURE sequencing simulation
- Simulate reads by masking regions across the genome (-excl): the coordinates for the regions to be excluded should be in a bed file structure.
- Added capture (-cap|--capture) simulations of regions solely containing variants provided by the vcf or bcf file (-vcf|-bcf)
- Added simulation of variants in LD (-linkage), by merging those regions in vcf which overlaps with each other after adding number of flanking (-fl) nucleotide to the variant position
- Added -name | --headername option to provide an individuals name when simulating variants from vcf files, with the name defined in vcf header
- Added an amplicon simulation mode: simulate sequencing errors, mutations or deamination on your empirical data, while allowing for file conversion, see NGSNGS - amplicon alteration
Next Generation Simulator for Next Generator Sequencing Data version 0.9.2.1
Next Generation Simulator for Next Generator Sequencing Data version 0.9.2.1
Usage
./ngsngs [options] -i <input_reference.fa> -r/-c <Number of reads or depth of coverage> -l/-lf/-ld <fixed length, length file or length distribution> -seq <SE/PE> -f <output format> -o <output name prefix>
Example
./ngsngs -i Test_Examples/Mycobacterium_leprae.fa.gz -r 100000 -t 2 -s 1 -lf Test_Examples/Size_dist_sampling.txt -seq SE -m b,0.024,0.36,0.68,0.0097 -q1 Test_Examples/AccFreqL150R1.txt -f bam -o MycoBactBamSEOut
./ngsngs -i Test_Examples/Mycobacterium_leprae.fa.gz -c 3 -t 2 -s 1 -l 100 -seq PE -ne -a1 AGATCGGAAGAGCACACGTCTGAACTCCAGTCACCGATTCGATCTCGTATGCCGTCTTCTGCTTG -a2 AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGTAGATCTCGGTGGTCGCCGTATCATTT -q1 Test_Examples/AccFreqL150R1.txt -q2 Test_Examples/AccFreqL150R2.txt -f fq -o MycoBactFqPEOut
./ngsngs -i Test_Examples/Mycobacterium_leprae.fa.gz -r 100000 -t 1 -s 1 -ld Pois,78 -seq SE -mf Test_Examples/MisincorpFile.txt -f fa -o MycoBactFaSEOut
./ngsngs -i Test_Examples/hg19MSub.fa -r 1000 -t 1 -s 100 -l 150 -seq SE -ne -vcf Test_Examples/ChrMtSubDeletionDiploid.vcf -id 0 -q1 Test_Examples/AccFreqL150R1.txt -chr MT -DumpVCF DeltionInfo -f fq -o MtDeletionOut
./ngsngs -i Test_Examples/hg19MSub.fa -r 100 -t 1 -s 1 -l 100 -seq SE -qs 40 -f fq -o CircularSimulation -sim circ
-h | --help: Print help page.
----- Required -----
-i | --input: Reference file in fasta format (.fa,.fasta) to sample reads.
Sequence reads:
-r | --reads: Number of reads to simulate, conflicts with -c option.
-c | --coverage: Depth of Coverage to simulate, conflics with -r option.
Fragment Length:
-l | --length: Fixed length of simulated fragments, conflicts with -lf & -ld option.
-lf | --lengthfile: CDF of a length distribution, conflicts with -l & -ld option.
-ld | --lengthdist: Discrete or continuous probability distributions, given their Probability density function, conflicts with -l & -lf option.
<Uni,Min,Max> Uniform distribution from a closed interval given a minimum and maximum positive integer, e.g. Uni,40,180.
<Norm,Mean,Variance> Normal Distribution, given a mean and variance, e.g. Norm,80,30.
<LogNorm,Mean,Variance> Log-normal Distribution, given a mean and variance, e.g. LogNorm,4,1.
<Pois,Rate> Poisson distribution, given a rate, e.g. Pois,165.
<Exp,Rate> Exponential distribution, given a rate, e.g. Exp,0.025.
<Gamma,Shape,Scale> Gamma distribution, given a shape and scale, e.g. Gam,20,2.
note: Depending on parameters, if the sampled length is below the lower limit (-ll) resampling is performed.
Output characteristics:
-seq | --sequencing: Simulate single-end or paired-end reads.
<SE||se||single||single-end> single-end.
<PE||pe||paired||paired-end> paired-end.
-f | --format: File format of the simulated output reads.
Nucletide sequence w. different compression levels.
<fa||fasta>
<fa.gz||fasta.gz>
Nucletide sequence with corresponding quality score w. different compression levels.
<fq||fastq>
<fq.gz||fastq.gz>
Sequence Alignment Map w. different compression levels.
<sam||bam||cram>
-o | --output: Prefix of output file name.
Format specific:
-q1 | --quality1: Read Quality profile for single-end reads (SE) or first read pair (PE) for fastq or sequence alignment map formats.
-q2 | --quality2: Read Quality profile for for second read pair (PE) for fastq or sequence alignment map formats.
-qs | --qualityscore: Fixed quality score, for both read pairs in fastq or sequence alignment map formats. It overwrites the quality profiles.
----- Optional -----
Simulation mode:
-circ | --circular: Circular simulations - Creating breakpoint reads crossing end coordinate of the reference genome for bacterial or mitochondrial simulation.
-chr | --chromosomes: Specific chromosomes from input reference file.
-incl | --include: Simulate sequencing reads solely from regions of interest within the input bed file.
-excl | --exclude: Simulate sequencing reads masking the regions within the input bed file.
-fl | --flanking: Fixed number of nucleotides flanking the regions of interest (-incl) in both ends, default = 30.
Reference Variations:
-mr | --mutationrate: Adding stochastic variations to the reference genome (-i) from a fixed mutation rate, conflicts with number of variations (-v), default = 0.0
-g | --generations: Adding stochastic variations to the reference genome (-i) according to the fixed mutation rate (-mr) across numerous generations, default = 1
-v | --variations: Adding a fixed number of stochastic variations to the reference genome (-i), conflicts with mutation rate (-mr), default = 0
Genetic Variations:
-bcf | -vcf: Variant Calling Format (.vcf) or binary format (.bcf)
-id | --indiv: Integer value (0 - index) for the number of a specific individual defined in bcf header from -vcf/-bcf input file, default = -1 (no individual selected).
e.g -id 0 First individual in the provided vcf file.
-name | --headername: The name of an individual defined in bcf header from -vcf/-bcf input file, conflicts with -id.
-DumpVCF: The prefix of an internally generated fasta file, containing the sequences representing the haplotypes with the variations from the provided
vcf file, for diploid individuals the fasta file contains two copies of the reference genome with the each allelic genotype.
-cap | --capture Simulates sequence reads solely from regions with variants provided by -vcf input, each variant is treated as an indepdent region
-linkage | --linkagedisequilibrium Simulates sequence reads from regions with variants provided by -vcf input, where varaiants in close proximity is treated as being in LD and simulated together
Stochastic Variations:
-indel: Input probabilities and lambda values for a geometric distribution randomly generating insertions and deletions of a random length.
<InsProb,DelProb,InsParam,DelParam>
Insertions and deletions -indel 0.05,0.1,0.1,0.2
Only Insertions -indel 0.05,0.0,0.1,0.0
Only Deletions -indel 0.0,0.5,0.0,0.9
-DumpIndel: The prefix of an internally generated text file, containing the the read id, number of indels, the number of indel operations saving the position
before and after and length of the indel, simulated read length before and after, see supplementary material for detailed example and description.
Postmortem damage (PMD) - Deamination:
-m | --model: Choice of deamination model.
<b,nv,Lambda,Delta_s,Delta_d> || <briggs,nv,Lambda,Delta_s,Delta_d> Parameters for the damage patterns using the Briggs model altered to suit modern day library preparation.
<b7,nv,Lambda,Delta_s,Delta_d> || <briggs07,nv,Lambda,Delta_s,Delta_d> Parameters for the damage patterns using the Briggs model 2007.
nv: Nick rate pr site.
Lambda: Geometric distribution parameter for overhang length.
Delta_s: PMD rate in single-strand regions.
Delta_d: PMD rate in double-strand regions.
e.g -m b,0.024,0.36,0.68,0.0097
-dup | --duplicates: Number of PCR duplicates, used in conjunction with briggs modern library prep -m <b,nv,Lambda,Delta_s,Delta_d>
<1,2,4>, Default = 1.
Nucleotide Alterations:
-mf | --mismatch: Nucleotide substitution frequency file.
-ne | --noerror: Disabling the nucleotide substitutions based on nucleotide qualities.
Read Specific:
-na | --noalign: Using the SAM output as a sequence containing without alignment information.
-cl | --cycle: Read cycle length, the maximum length of sequence reads, if not provided the cycle length will be inferred from quality profiles (q1,q2).
-ll | --lowerlimit: fragment length limit, default = 30. The minimum fragment length for deamination is 30, so simulated fragments below will be fixed at 30.
-bl | --bufferlength: Buffer length for generated sequence reads stored in the output files, default = 30000000.
-a1 | --adapter1: Adapter sequence to add for simulated reads (SE) or first read pair (PE).
e.g. Illumina TruSeq Adapter 1: AGATCGGAAGAGCACACGTCTGAACTCCAGTCACCGATTCGATCTCGTATGCCGTCTTCTGCTTG
-a2 | --adapter2: Adapter sequence to add for second read pair (PE).
e.g. Illumina TruSeq Adapter 2: AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGTAGATCTCGGTGGTCGCCGTATCATTT
-p | --poly: Create Poly(X) tails for reads, containing adapters with lengths below the inferred readcycle length.
e.g -p G or -p A
Simulation Specific:
-t | --threads: Number of sampling threads, default = 1.
-t2 | --threads2: Number of compression threads, default = 0.
-s | --seed: Random seed, default = current calendar time (s).
-rng | --rand: Pseudo-random number generator, OS specific
<0,1,2,3>
0 : drand48_r, default for linux or unix, not available for MacOS.
1 : std::uniform_int_distribution
2 : rand_r
3 : erand48, default for MacOS.Next Generation Simulator for Next Generator Sequencing Data Amplicon simulation - adding nucleotide alterations to existing empirical data
Usage
./amplicon [options] --amplicon <InputSequenceFile> --output <AmpliconOutput>
-h | --help: Print help page.
Required:
-a | --amplicon: Input sequence file in .fq format.
-o | --output: Prefix for the output sequence file with nucleotide alterations.
Optional:
-f | --format: File format of the simulated amplicon file which can be used for file conversion
Nucletide sequence w. different compression levels.
<fa||fasta>
<fa.gz||fasta.gz>
Nucletide sequence with corresponding quality score w. different compression levels.
<fq||fastq>
<fq.gz||fastq.gz>
Sequence Alignment Map w. different compression levels.
<sam||bam>
-s | --seed: Random seed, default = current calendar time (s).
-rng | --rand: Pseudo-random number generator, OS specific
<0,1,2,3>
0 : drand48_r, default for linux or unix, not available for MacOS.
1 : std::uniform_int_distribution
2 : rand_r
3 : erand48, default for MacOS.
Sequencing errors:
-q | --quality: Read Quality profile to simulate nucleotide quality scores for fasta input file to generate fastq output.
-qs | --qualityscore: Simulate a fixed nucleotide quality score for fasta input file to generate fastq output.
-ne | --noerror: Disabling the sequencing error substitutions based on nucleotide qualities from the provided quality profile -q or fixed score -qs.
Postmortem damage (PMD) - Deamination:
-m | --model: Deamination model.
<b,nv,Lambda,Delta_s,Delta_d> || <briggs,nv,Lambda,Delta_s,Delta_d> Parameters for the damage patterns using the Briggs model.
nv: Nick rate per site.
Lambda: Geometric distribution parameter for overhang length.
Delta_s: PMD rate in single-strand regions.
Delta_d: PMD rate in double-strand regions.
e.g -m b,0.024,0.36,0.68,0.0097
Stochastic Variations:
-indel: Input probabilities and lambda values for a geometric distribution randomly generating insertions and deletions of a random length.
<InsProb,DelProb,InsParam,DelParam>
Indels -indel 0.05,0.1,0.1,0.2
Insertions -indel 0.05,0.0,0.1,0.0
Deletions -indel 0.0,0.5,0.0,0.9
Nucleotide Alterations:
-mf | --mismatch: Nucleotide substitution frequency file.
Example
./amplicon --amplicon Test_Examples/Amplicon_in.fq -m b,0.024,0.36,0.68,0.0097 --format fq.gz --output Amplicon_deamination
./amplicon --amplicon Test_Examples/Amplicon_in.fq -indel 0.05,0.1,0.1,0.2 --output Amplicon_indel
./amplicon --amplicon Test_Examples/Amplicon_in.fa --format fq -q Test_Examples/AccFreqL150R1.txt --output Amplicon_seqerr
./amplicon --amplicon Test_Examples/Amplicon_in.bam -mf ../Test_Examples/MisincorpFile.txt --format fa.gz --output Amplicon_deamination
Examples of which parameters to include depending on the desired simulations
simulate 1000 reads (-r) from human hg19.fa (-i), generate compressed fq.gz (-f), single end (-seq), make program use one threads (-t)
./ngsngs -i hg19.fa -r 1000 -f fq.gz -l 100 -seq SE -t 1 -q1 Test_Examples/AccFreqL150R1.txt -o HgSimsimulate 1000 reads (-r) from human hg19.fa (-i), generate fq (-f), single end (-seq), with a fixed quality score (-qs) and lower fragment limit of 50 (-ll)
./ngsngs -i hg19.fa -r 1000 -f fq -lf Test_Examples/Size_dist_sampling.txt -ll 50 -seq SE -t 1 -qs 40 -o HgSimgenerate bam (-f), paired end (-seq), fragment length from a normal distribution mean:350,sd:20 (-ld) but fixed cycle length (-cl) with two independent quality profiles (-q1,-q2)
./ngsngs -i hg19.fa -r 1000 -f bam -ld norm,350,20 -cl 100 -seq PE -t 1 -q1 Test_Examples/AccFreqL150R1.txt -q2 Test_Examples/AccFreqL150R2.txt -o HgSimDisable platform specific errors (-ne), adding deamination pattern with Briggs 2007 model (-m b7,...), with ancient fragment length distribution (-lf), using seed 4 (-s)
./ngsngs -i hg19.fa -r 1000 -f fq -s 4 -ne -lf Test_Examples/Size_dist_sampling.txt -seq SE -m b7,0.024,0.36,0.68,0.0097 -q1 Test_Examples/AccFreqL150R1.txt -o HgSimPaired end reads, inferred cycle length from the distribution of quality profile (-q1) to be 150, fragment length (-l) 400, inner distance of 100 (400-150*2), while adding adapters
./ngsngs -i hg19.fa -r 1000 -f fq -l 400 -seq PE -q1 Test_Examples/AccFreqL150R1.txt -q2 Test_Examples/AccFreqL150R1.txt -a1 AGATCGGAAGAGCACACGTCTGAACTCCAGTCACCGATTCGATCTCGTATGCCGTCTTCTGCTTG -a2 AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGTAGATCTCGGTGGTCGCCGTATCATTT -o HgSim./ngsngs -i hg19.fa -r 1000 -f fq -l 80 -seq SE -q1 Test_Examples/AccFreqL150R1.txt -a1 AGATCGGAAGAGCACACGTCTGAACTCCAGTCACCGATTCGATCTCGTATGCCGTCTTCTGCTTG -p G -o HgSimNB! the adapter sequences are only concatenated to the reads, if the inferred cycle length from quality profiles is greater than the fragment length, the poly-X tail is only added if the sequence with adapter length is still below the cycle length (Cycle length - fragment length - adapter length = 150 - 80 - 65 = 5).
All formats shares a similar read ID structure
T<ThreadNumber>_RID<RandomID>_S<Read1StrandInfo>_<Chromosome>:<Start>-<End>
_length:<Fragmentlength>_<modVal1Val2Val3Val4> F<FragmentNumber> R<PairNumber>e.g. @T0_RID49_S0_NZ_CP029543.1:2236795-2236942_length:148_mod1000 F0 R1
S0 is the forward strand and S1 is the reverse strand, mod1000 equals read with deamination, F0 signifies the first fragment out of 4 possible PCR duplicates, R1 indicate the sequence is read 1 (See supplementary material for detailed description).
- Val1 signifies deamination of the simulated fragment, 0 = no deamination, 1 = deamination.
- Val2 signifies mismatches from file (-mf) of the simulated fragment, 0 = no mismatch, 1 = 5' mismatch, 2 = 3' mismatch, 3 = mismatch in both ends.
- Val3 signifies stochastic structural variations in the sequenced reads, 0 = no variation, 1 = insertions, 2 = deletions, 3 = insertions and deletions.
- Val4 signifies sequencing error (dependent on -q1,-q2 or -qs) in the sequenced reads, 0 = no sequencing error, 1 = sequencing error.
Simulating a .fq or .sam format requires a provided nucleotide quality profile (-q1, -q2) with one example (Test_Examples/AccFreqL150R1.txt) and its structure:
Line1: 3 7 16 23 28 34 38 41
Line2: 0.501187 0.199526 0.025119 0.005012 0.001585 0.000398 0.000158 0.000079
Line3: 5.848056e-07 5.853904e-04 2.188810e-02 2.653555e-02 1.208910e-01 1.000000e+00 0.000000e+00 0.000000
Line4: 3.960783e-07 4.626195e-04 1.261628e-02 1.813564e-02 7.776444e-02 1.000000e+00 0.000000e+00 0.000000
Line5: 3.450651e-07 8.071072e-04 5.794989e-02 8.024696e-02 1.755377e-01 1.000000e+00 0.000000e+00 0.000000
.
.
.
Line750: 1.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00
Line751: 1.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00
Line752: 1.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00- The first two lines signify the possible quality scores to be simulated and their corresponding error probability.
- The remaining lines are separated into 5 equal regions, i.e 750/5 -> 150. Signifying the number of positions for which a read can have simulated nucleotide quality scores. The first 150 lines -> A, next T, then G, then C, and finally N.
- When quality profiles are provided (-q1,-q2), NGSNGS infers the cycle length given the number of positions with corresponding quality scores (number of lines) and creates a read length limitation (e.g. (752-2)/5 = 150) independent of the fragment length options (-l,-lf,-ld).
- To simulate .fastq outputs, the quality profiles need to be provided. For Sequence Alignment Map formats, if no quality profile has been provided, then all the quality scores in the nucleotide quality string will be the lowest quality.
- From the error probability depending on the simulated quality score, sequencing errors will be simulated by equally substituting between the remaining three nucleotides. When providing the '-ne' option sequencing error substitution is disabled.
This misincorporation file (-mf) represents the type-specific probabilities of any of the bases transitioning to any other nucleotide or not transitioning for every cycle or position of the sequence. The structure (Test_Examples/MisincorpFile.txt) is similar to that of the nucleotide quality profiles, with the number of bases for which nucleotide substitutions can occur being inferred from the dimension.
A T G C
Line1: 0.865434 0.888339 0.953086 1.000000
Line2: 0.882001 0.894563 0.979380 1.000000
Line3: 0.915281 0.932903 0.983310 1.000000
.
.
.
Line118: 0.004576 0.023394 0.029278 1.000000
Line119: 0.002286 0.017742 0.021349 1.000000
Line120: 0.003241 0.015273 0.019467 1.000000- The substitution pattern from the misincorporation file, represent the substitution of all four nucleotide with the first half being from from both the 5’ termini and the second half being the 3’ termini of a given fragment.
- With a dimension of 120 lines, the first 60 represent substitution frequencies of the first 15 positions within the read given the nucleotide belonging to A,T,G or C and the latter half being the last 15 nucleotides.
The CDF of the fragment lengths distribution (-lf) of the DNA sequences from which the sequence reads are generated.
35 0.00540914
36 0.01326621
37 0.02248544
38 0.03442894
39 0.04907704
.
.
.
187 0.9997630062
188 0.9998814542
189 0.999920937
190 0.9999406784
191 1- Given the above nucleotide quality profile, those reads with a fragment length above the inferred the cycle length (150 bp), will have an discrepancy between the read id "length:<Fragmentlength>" and the simulated output sequence length.
As of release v0.9.2.1 NGSNGS has the following restrictions on fragment length simulations (-l,-ld,-ld) and the modifications which can be performed on the fragment. These will be addressed in future releases.
- The upper limit of fragment lengths is 10000
- When simulating deamination fragments below 30 nucleotides will not be deamination
- When simulating deamination fragments above 1024 will not be deaminated in the 3' end with the complementary G>A substitution
Bibtex citation
@article{10.1093/bioinformatics/btad041,
author = {Henriksen, Rasmus Amund and Zhao, Lei and Korneliussen, Thorfinn Sand},
title = "{NGSNGS: Next generation simulator for next generation sequencing data}",
journal = {Bioinformatics},
year = {2023},
month = {01},
issn = {1367-4811},
doi = {10.1093/bioinformatics/btad041},
url = {https://doi.org/10.1093/bioinformatics/btad041},
note = {btad041},
eprint = {https://academic.oup.com/bioinformatics/advance-article-pdf/doi/10.1093/bioinformatics/btad041/48800233/btad041.pdf},
}
