SVDB is a toolkit for constructing and querying structural variant databases. The databases are constructed using the output vcf files from structural variant callers such as TIDDIT, Manta, Fermikit or Delly.
Additionally, the thousand genomes structural variant calls may also be used as a database: ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/phase3/integrated_sv_map/
or the swegen SVDB:
or the GNOMAD SV:
https://storage.googleapis.com/gnomad-public/papers/2019-sv/gnomad_v2_sv.sites.vcf.gz
external databses are run like this:
zgrep -E "#|POPM" gnomad_v2_sv.sites.vcf.gz > popmax_sv_gnomad.vcf
svdb --query --db popmax_sv_gnomad.vcf --query_vcf P2109_110.clean.dedup.recal_FindSV.vcf --frequency_tag AF --hit_tag AN
here the AF and AN are the allele frequency tags of the database, the AF is a float, and AN is an integer. These tags will be a dded to the matching entries of the query.
NOTE: AF and AN needs cannot be comma separated values, only single integer or float values are supported!
Dependencies: SVDB has been tested on python 2.7.11, and requires numpy. SVDB is installed using the following command
pip install -e .
SVDB consists of modules that are used to build, query, export, and analyse structural variant databases. These are the modules:
Build: This module is used to construct structural variant databases from vcf files. The database may then be queried to compute the frequency of structural variants, or exported into a vcf file. These are the commands used to construct a structural variation database:
print a help message
svdb --build --help
Construct a database, from a set of vcf files:
svdb --build --vcf sample1.vcf sample2.vcf sample3.vcf
construct a database from vcf files stored in a folder
svdb --build --folder SV_analysis_folder/
optional arguments:
-h, --help show this help message and exit
--files [FILES [FILES ...]] create a db using the specified vcf files(cannot be
used with --folder)
--folder FOLDER create a db using all the vcf files in the folders
--prefix PREFIX the prefix of the output file, default = SVDB
Export: this module is used to export the variants of the SVDB sqlite database. The variants of the sqlite svdb database is clustered using one out of three algorihms, overlap, DBSCAN, or CIPOS.
print a help message
svdb --export --help
Export the variants of the database database.db:
svdb --export --db database.db
optional arguments:
--no_merge skip the merging of variants, print all variants in the db to a vcf file
--ci overides overlap and bnd_distance,determine hits based on the confidence interval of the position of the variants(0 if no CIPOS or CIEND is vailable)
--bnd_distance BND_DISTANCE the maximum distance between two similar precise breakpoints(default = 2500)
--overlap OVERLAP the overlap required to merge two events(0 means anything that touches will be merged, 1 means that two events must be identical to be merged), default = 0.8
--DBSCAN use dbscan to cluster the variants, overides the overlap based clustering algoritm
--epsilon EPSILON used together with --DBSCAN; sets the epsilon paramter(default = 500bp)
--min_pts MIN_PTS used together with 1--DBSCAN; sets the min_pts parameter(default = 2
--prefix PREFIX the prefix of the output file, default = same as input
--memory load the database into memory: increases the memory requirements, but lowers the time consumption
Hist: This module is used to compare structural variant vcf files, either by generating a similarity matrix, or by creating histograms of the efficency of databases of different sizes(based on input vcf files):
print a help message
svdb --hist --help
Create histograms of different sizes, and compute their efficiency:
svdb --hist --sample_hist -folder input
Create a similarity matrix of the selected sampes:
svdb --hist --similarity_matrix -folder input
optional arguments:
-h, --help show this help message and exit
--files [FILES [FILES ...]] input vcf files(cannot be used with folder)
--k [K [K ...]] the sizes of the sampled databases
default = n=10*i < samples(used with sample_hist)
--n N the number of iterations,default=100(used with sample_hist)
--bnd_distance BND_DISTANCE the maximum distance between two similar precise
breakpoints(default = 10000)
--overlap OVERLAP the overlap required to merge two events(0 means
anything that touches will be merged, 1 means that two
events must be identical to be merged), default = 0.6
--ci overides overlap and bnd_distance,determine hits based
on the confidence interval of the position of the
variants(0 if no CIPOS or CIEND is vailable)
Query: The query module is used to query a structural variant database. Typically a database is constructed using the build module. However, since this module utilize the genotype field of the sructural variant database vcf to compute the frequency of structural variants, a wide range of files could be used as database. The query module requires a query vcf, as well as a database file(either multisample vcf or SVDB sqlite database):
print a help message
svdb --query --help
Query a structural variant database, using a vcf file as query:
svdb --query --query_vcf patient1.vcf --db control_db.vcf
The vcf may be a exported SVDB database or a multismple vcf. The frequencies used for each variant is computed from the format fields of the vcf.
The SVDB sqlite database may also be used for querying:
svdb --query --query_vcf patient1.vcf --sqdb control_db.db
optional arguments:
-h, --help show this help message and exit
--db DB path to a SVDB db vcf
--sqdb SQDB path to a SVDB sqlite db
--hit_tag HIT_TAG the tag used to describe the number of hits within the info field of the output vcf(default=OCC)
--frequency_tag FREQUENCY_TAG the tag used to describe the frequency of the variant(defualt=FRQ)
--prefix PREFIX the prefix of the output file, default = print to stdout --bnd_distance BND_DISTANCE the maximum distance between the breakpoints of two variantsbreakpoints(default = 10000)
--overlap OVERLAP the overlap threshold for deciding if two variants are similar(0 means anything that touches will be merged, 1 means that two events must be identical to be merged), default = 0.6
--memory load the database into memory: increases the memory requirements, but lowers the time consumption(may only be used with sqdb)
--no_var count overlaping variants of different type as hits in the db
--invert invert the sorting order so that high frequency samples are present on top of the output vcf
--ci overides overlap and bnd_distance,determine hits based on the confidence interval of the position fo the variants(0 if no CIPOS or CIEND is vailable)
Merge: The merge module merges variants within one or more vcf files. This could be used to either merge the output of multiple callers, or to merge variants that are called multiple times due to noise or some other error:
print a help message:
python SVDB.py --merge --help
merge vcf files:
svdb --merge --vcf patient1_lumpy.vcf patient1_cnvnator.vcf patient1_TIDDIT.vcf > patient1_merged_callers.vcf
Similar variants will be merged, and presented according to the order of the input vcf files. I.e If lumpy and cnvnator calls the same variant in the top example,
the variant will be printed as the lumpy call. In most cases, the order should be set according to the accuracy or detail of the info field of the different callers.
The order could also be set using the --priority flag:
svdb --merge --vcf patient1_lumpy.vcf:one patient1_cnvnator.vcf:2 patient1_TIDDIT.vcf:tiddit --priority tiddit,2,one > patient1_merged_callers.vcf
In this example, tiddit will have the highest order, cnvnator second etc.
optional arguments:
-h, --help show this help message and exit
--bnd_distance BND_DISTANCE the maximum distance between two similar precise breakpoints
(default = 10000)
--overlap OVERLAP the overlap required to merge two events(0 means
anything that touches will be merged, 1 means that two
events must be identical to be merged), default = 0.6
--priority prioritise the input vcf files
--ci overides overlap and bnd_distance,determine hits based
on the confidence interval of the position fo the
variants(0 if no CIPOS or CIEND is vailable)
--no_intra no merging of variants within the same vcf
--no_var variants of different type will be merged
--pass_only merge only variants labeled PASS