cuteSV

Getting Start

                                               __________    __       __
                                              |   ____   |  |  |     |  |
                          _                   |  |    |__|  |  |     |  |
 _______    _     _   ___| |___     ______    |  |          |  |     |  |
|  ___  |  | |   | | |___   ___|   / ____ \   |  |_______   |  |     |  |
| |   |_|  | |   | |     | |      / /____\ \  |_______   |  |  |     |  |
| |        | |   | |     | |      | _______|   __     |  |  \  \     /  /
| |    _   | |   | |     | |  _   | |     _   |  |    |  |   \  \   /  /
| |___| |  | |___| |     | |_| |  \ \____/ |  |  |____|  |    \  \_/  /
|_______|  |_______|     |_____|   \______/   |__________|     \_____/

Installation

$ pip install cuteSV
or
$ conda install -c bioconda cutesv
or 
$ git clone https://github.com/tjiangHIT/cuteSV.git && cd cuteSV/ && python setup.py install 

Introduction

Long-read sequencing enables the comprehensive discovery of structural variations (SVs). However, it is still non-trivial to achieve high sensitivity and performance simultaneously due to the complex SV characteristics implied by noisy long reads. Therefore, we propose cuteSV, a sensitive, fast and scalable long-read-based SV detection approach. cuteSV uses tailored methods to collect the signatures of various types of SVs and employs a clustering-and-refinement method to analyze the signatures to implement sensitive SV detection. Benchmarks on real Pacific Biosciences (PacBio) and Oxford Nanopore Technology (ONT) datasets demonstrate that cuteSV has better yields and scalability than state-of-the-art tools.

The benchmark results of cuteSV on the HG002 human sample are below:

BTW, we used Truvari to calculate the recall, precision, and f-measure. For more detailed implementation of SV benchmarks, we show an example here.

Dependence

1. python3
2. pysam
3. Biopython
4. cigar
5. numpy
6. pyvcf

Usage

cuteSV <sorted.bam> <reference.fa> <output.vcf> <work_dir>

Suggestions

> For PacBio CLR data:
	--max_cluster_bias_INS		100
	--diff_ratio_merging_INS	0.3
	--max_cluster_bias_DEL	200
	--diff_ratio_merging_DEL	0.5

> For PacBio CCS(HIFI) data:
	--max_cluster_bias_INS		1000
	--diff_ratio_merging_INS	0.9
	--max_cluster_bias_DEL	1000
	--diff_ratio_merging_DEL	0.5

> For ONT data:
	--max_cluster_bias_INS		100
	--diff_ratio_merging_INS	0.3
	--max_cluster_bias_DEL	100
	--diff_ratio_merging_DEL	0.3

Datasets generated from cuteSV

We provided the SV callsets of the HG002 human sample produced by cuteSV form three different long-read sequencing platforms (i.e. PacBio CLR, PacBio CCS, and ONT PromethION).

You can download them at:

Please cite the manuscript of cuteSV before using these callsets.

Changelog

cutesv (v1.0.10):
1. Fix a bug leading to calculate wrong TRA positions.
2. Add a file format conversion script that enable to transfer the vcf file to bedpe file. 
3. Involve several clustering-and-refinement strategies in force calling function.
4. Assessed the performance of force calling with Giab HG002 sample datasets (including CLR, CCS, and ONT platforms).

cutesv (v1.0.9):
1. Change 0-based pos into 1-based pos in DUP in order to support bcftools conversion.
2. Correct REF and ALT fields. Adjust END value of INS to make it equal to the value of POS.
3. Improve the description of errors.
4. Add usegalaxy.eu badge.
5. Remove CHR2 and the corresponding END position on the BND call.
6. Skip generating empty signature file and rewrite the job schedule.
7. Add force calling function and enable cuteSV to perform population-based SV calling.
8. Fix several minor bugs.

cutesv (v1.0.8):
1. Rewirte the function of ins/del signatures clustering.
2. Update the recommandation parameters for different sequencing datasets.
3. Replace <DEL>/<INS> with its variant allele sequence, which needs the reference genome sequence as input.
4. Fix several bugs.

cuteSV (v1.0.7):
1. Add read name list for each SV call.
2. Fix several descriptions in VCF header field.

cuteSV (v1.0.6):
1.Improvement of genotyping by calculation of likelihood.
2.Add variant quality value, phred-scaled genotype likelihood and genotype quality in order to filter false positive SV or quality control.
3.Add --gt_round parameter to control the number of read scans.
4.Add variant strand of DEL/DUP/INV.
5.Fix several bugs.

cuteSV (v1.0.5):
1.Add new options for specificly setting the threshold of deletion/insertion signals merging in the same read. The default parameters are 0 bp for deletion and 100 bp for insertion.
2.Remove parameter --merge_threshold.
3.Fix bugs in inversion and translocation calling.
4.Add new option for specificly setting the maximum size of SV to be discovered. The default value is 100,000 bp. 


cuteSV (v1.0.4):
1.Add a new option for specificly setting the threshold of SV signals merging in the same read. The default parameter is 500 bp. You can reduce it for high-quality sequencing datasets like PacBio HiFi (CCS).
2.Make the genotyping function optional.
3.Enable users to set the threshold of SV allele frequency of homozygous/heterozygous.
4.Update the description of recommendation parameters in processing ONT data.

cuteSV (v1.0.3):
1.Refine the genotyping model.
2.Adjust the threshold value of heterozygosis alleles.

cuteSV (v1.0.2):
1.Improve the genotyping performance and enable it to be default option.
2.Make the description of parameters better.
3.Modify the header description of vcf file.
4.Add two new indicators, i.e., BREAKPOINT_STD and SVLEN_STD, to further characterise deletion and insertion.
5.Remove a few redundant functions which will reduce code readability.

Citation

Jiang T et al. Long-read-based human genomic structural variation detection with cuteSV. Genome Biol 21, 189 (2020). https://doi.org/10.1186/s13059-020-02107-y

Contact

For advising, bug reporting and requiring help, please post on Github Issue or contact tjiang@hit.edu.cn.