OncodriveCLUSTL is a clustering method to identify cancer drivers
Project description
OncodriveCLUSTL
OncodriveCLUSTL is a sequence-based clustering method to identify significant clustering signals in nucleotide sequence.
One of the main goals of cancer research is the identification of the genomic elements that drive tumorigenesis. OncodriveCLUSTL is a new nucleotide sequence-based clustering algorithm to detect significant clustering signals across genomic regions. OncodriveCLUSTL is based on a local background model derived from the nucleotide context mutational probabilities of the cohort under study. Our method is able to identify well-known cancer drivers in coding regions and it can be applied to non-coding regions and non-human data.
License
OncodriveCLUSTL is available to the general public subject to certain conditions described in its LICENSE.
Installation
OncodriveCLUSTL depends on Python 3.5 or above. We recommend to install it using the Anaconda Python distribution:
conda install -c bbglab oncodriveclustl
OncodriveCLUSTL can also be installed using pip:
pip install oncodriveclustl
You can obtain the latest code from the repository and install it for development with pip:
git clone git@bitbucket.org:bbglab/oncodriveclustl.git
cd oncodriveclustl
python -m venv .venv
.venv/bin/pip install -e .
source .venv/bin/activate
oncodriveclustl --help
[!NOTE] The first time that you run OncodriveCLUSTL with a given reference genome, it will download it from our servers. By default the downloaded datasets go to
~/.bgdata
. If you want to move these datasets to another folder you have to define the system environment variableBGDATA_LOCAL
with an export command.
[!NOTE] If you install a modern build tool like uv, you can simply do this:
git clone https://github.com/bbglab/oncodriveclustl.git cd oncodriveclustl uv run oncodriveclustl --help
Input data
OncodriveCLUSTL only requires two main inputs, the mutations file and the annotations file.
Mutations file
TSV file containing SNVs (substitutions) mapped to a reference genome (e.g., human hg19 or mouse c3h). If other mutation types are present (insertions, deletions, double base substitutions, etc), they will be filtered out during the analysis. This file must contain, at least, the following 5 columns with header:
- CHROMOSOME: 1, 2,..., X, Y
- POSITION: Integer indicating the position of the mutation
- REF: Reference nucleotide
- ALT: Alternate nucleotide
- SAMPLE: Identifier of the sample
Additional columns are:
- CANCER_TYPE: Type of tumor. When specified, OncodriveCLUSTL will calculate one mutational profile for each cancer type and mutations will be randomized accordingly.
- SIGNATURE: User-defined group to compute k-mer nucleotide mutational probabilities. When specified, OncodriveCLUSTL will calculate one mutational profile for each group and will randomize each mutation accordingly.
[!NOTE] OncodriveCLUSTL assumes all SNVs are mapped to the positive strand.
[!WARNING] When using the
--signature-group
option, please check that the number of mutations per group is sufficient for an accurate signatures calculation.
Annotations file
TSV file containing the coordinates of genomic elements (GEs). This file must contain, at least, the following 5 columns with header:
- CHROMOSOME: 1, 2,..., X, Y
- START: Starting position of the genomic region
- END: Final position of the genomic region
- ELEMENT: Identifier of the GE
- SYMBOL: Symbol of the GE. OncodriveCLUSTL will analyze GEs as SYMBOL + ELEMENT.
Additional columns are:
- STRAND: Strand of the GE coordinates ("+" or "-").
[!WARNING] Coordinates of a given GE cannot overlap.
You can check the input formats in the files provided in the example.
If you have a VCF file or directory of VCF files containing somatic mutations, you can run our VCF parser to obtain a tabular file compatible with OncodriveCLUSTL input format::
parse_vcf -i [INPUT_DIRECTORY] -o [OUTPUT_FILE]
Please, check parsers/vcf.py module for more details.
If you would like to run OncodriveCLUSTL using a per-calculated signature or mutational profile, you need to provide a dictionary containing the reference k-mer to alternate mutational probabilities in JSON format:
{
"my_dataset": {
"GCA>G": 0.02424271083094251,
"AGC>A": 0.023005887103025254,
"ACG>T": 0.037613802858829135,
"CGA>C": 0.10691031051670515,
"GAC>G": 0.017846071811001615,
"TTC>A": 0.024003748061871697,
"CTT>G": 0.024149863672267024,
"GGA>T": 0.011178562948734577,
"AGG>C": 0.010654720767868876,
"GGG>C": 0.012031686292218055,
"CAA>T": 0.014478959792844522,
"TGA>A": 0.01255651801972085,
"GGA>A": 0.011178562948734577,
"CGA>A": 0.03563677017223505,
"TCC>T": 0.011158347971568658,
"GCC>A": 0.010952316565906438,
// ...
}
}
OncodriveCLUSTL requires non-collapsed k-mer probabilities (192 for tri-nucleotides, 3072 for penta-nucleotides).
Output data
OncodriveCLUSTL generates three output files:
Elements results file ('elements_results.txt')
TSV file containing results of the analyzed elements:
- SYMBOL: GE symbol #. ENSID: GE ID #. CGC: True if GE in the COSMIC Cancer Gene Census (CGC) list (Sondka et al., 2018)
- CHROMOSOME: 1, 2,..., X, Y
- STRAND: Strand of the GE ("+" or "-")
- LENGTH: length (bp) of the GE
- TOTAL_MUT: total substitutions observed in the GE
- CLUSTERED_MUT: number of substitutions in a cluster
- CLUSTERS: number of clusters
- SIM_CLUSTERS: number of simulated clusters
- SCORE: GE score
- P_EMPIRICAL: empirical p-value of the GE
- Q_EMPIRICAL: empirical q-value of the GE
- P_ANALYTICAL: analytical p-value of the GE
- Q_ANALYTICAL: analytical q-value of the GE
- P_TOPCLUSTER: analytical p-value of the cluster with highest cluster score
- Q_TOPCLUSTER: analytical q-value of the cluster with highest cluster score
Clusters results file ('clusters_results.tsv').
TSV file containing results of the clusters observed in the analyzed elements:
- RANK: Position of the GE in the list of
- SYMBOL: GE symbol
- ENSID: GE ID
- CGC: True if GE in the CGC list
- CHROMOSOME: 1, 2,..., X, Y
- STRAND: Strand of the GE ("+" or "-")
- COORDINATES: genomic coordinates of the cluster. It can be 'coord1,coord2' for clusters inside a single region or 'coord1,coord2;coord3,coord4' for those spanning regions (--concatenate flag)
- MAX_COORD: genomic position with the highest smoothing score inside the cluster
- WIDTH: cluster's width (pb)
- N_MUT: number of substitutions in the cluster
- N_SAMPLES: number of samples with a mutation in the cluster
- FRA_UNIQ_SAMPLES: proportion of unique samples mutated in the cluster out of the total of mutations in the cluster
- SCORE: cluster score
- P: analytical p-value of the cluster
Log file ('results.log')
TXT file containing OncodriveCLUSTL's run information.
Usage
OncodriveCLUSTL is meant to be used through the command line.
Usage: oncodriveclustl [OPTIONS]
Options:
-i, --input-file PATH File containing somatic mutations
[required]
-r, --regions-file PATH File with the genomic regions to analyze
[required]
-o, --output-directory TEXT Output directory to be created [required]
-sig, --input-signature PATH File containing input context based
mutational probabilities (signature)
-ef, --elements-file PATH File with the symbols of the elements to
analyze
-e, --elements TEXT Symbol of the element(s) to analyze
-g, --genome [hg38|hg19|mm10|c3h|car|cast|f344]
Genome to use
-emut, --element-mutations INTEGER
Cutoff of element mutations. Default is 2
-cmut, --cluster-mutations INTEGER
Cutoff of cluster mutations. Default is 2
-sw, --smooth-window INTEGER RANGE
Smoothing window. Default is 11 [3<=x<=101]
-cw, --cluster-window INTEGER RANGE
Cluster window. Default is 11 [3<=x<=101]
-kmer, --kmer [3|5] K-mer nucleotide context
-n, --n-simulations INTEGER number of simulations. Default is 1000
-sim, --simulation-mode [mutation_centered|region_restricted]
Simulation mode
-simw, --simulation-window INTEGER RANGE
Simulation window. Default is 31
[19<=x<=101]
-sigcalc, --signature-calculation [frequencies|region_normalized]
Signature calculation: mutation frequencies
(default) or k-mer mutation counts
normalized by k-mer region counts
-siggroup, --signature-group [SIGNATURE|SAMPLE|CANCER_TYPE]
Header of the column to group signatures
calculation
-c, --cores INTEGER RANGE Number of cores to use in the computation.
By default it will use all the available
cores. [1<=x<=10]
--seed INTEGER Seed to use in the simulations
--log-level [debug|info|warning|error|critical]
Verbosity of the logger
--concatenate Calculate clustering on concatenated genomic
regions (e.g., exons in coding sequences)
--clustplot Generate a needle plot with clusters for an
element
--qqplot Generate a quantile-quantile (Q-Q) plot for
a dataset
--gzip Gzip compress files
-h, --help Show this message and exit.
[!NOTE] When using simulation mode 'mutation_centered', simulation windows can be simulated outside the GE.
[!NOTE] When using
--signature-calculation region_normalized
, k-mer mutation counts will be normalized by k-mer nucleotide counts in the genomic regions provided as input (--regions-file
).
Run the example
If you run OncodriveCLUSTL from the source code, you can run an example of TCGA pancreatic adenocarcinomas (Ellrott et al. 2018) for coding regions (Mularoni et al., 2016) using 1000 simulations. First you need to download the example folder. Then you run OncodriveCLUSTL with default mode and parameters as:
oncodriveclustl -i example/PAAD.tsv.gz -r example/cds.hg19.regions.gz -o example/output
The results will be saved in a folder named output
.
You can compute a more sophisticated analysis using non-default parameters and generate a quantile-quantile plot by typing:
oncodriveclustl -i example/PAAD.tsv.gz -r example/cds.hg19.regions.gz -o example/output -sw 15 -cw 15 -simw 35 -sim region_restricted --concatenate --qqplot
If you want to run a specific GE and generate a plot its observed clusters, you can type::
oncodriveclustl -i example/PAAD.tsv.gz -r example/cds.hg19.regions.gz -o example/output -sw 15 -cw 15 -simw 35 -sim region_restricted --concatenate --clustplot -e KRAS
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file oncodriveclustl-1.2.0.tar.gz
.
File metadata
- Download URL: oncodriveclustl-1.2.0.tar.gz
- Upload date:
- Size: 2.3 MB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: uv/0.5.8
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 |
6ede4f8b2dc7550fa0ef407ccd13ed8c4427e9e5d6184742e36a2b7d879ae64c
|
|
MD5 |
3d995ce0f6f8469cb4e5ce643990cf84
|
|
BLAKE2b-256 |
cc2f681f2d18296f0c9953448fbb06a76a9873ce040f953a15ca220e4ad4dd48
|
File details
Details for the file oncodriveclustl-1.2.0-py3-none-any.whl
.
File metadata
- Download URL: oncodriveclustl-1.2.0-py3-none-any.whl
- Upload date:
- Size: 38.8 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: uv/0.5.8
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 |
bdcf096290f11169737cbef10505acbb7e48aa7745cc796d996d5ee6b19be947
|
|
MD5 |
006d4fa9eac3988f99d0d20859a0c7de
|
|
BLAKE2b-256 |
9c3f0a2b23357e7d5a13515c965552d42c8a0fdc2d3d9e7f57ada59bb5498d84
|