The easiest way to run BaRDIC is to launch it from the command line. There will be an option for python API in the future.
BaRDIC needs only 4 files as input:
| File | Explanation | Extension |
|---|---|---|
| dnaparts | DNA parts of RNA-DNA contacts as BED6 file with names of RNAs in the name field. score and strand fields are not relevant. |
bed |
| annotation | Genome annotation that contains coordinates of RNAs genes in BED6 format. | bed |
| chromsizes | Either UCSC genome code name or two-column headerless tab-delimeted file with chromosome name and chromosome length in each row. | txt or genome code name |
| bg_rnas | List of RNAs names which DNA parts of contacts to use for background estimation, one RNA name per line. | txt |
BaRDIC is a multi-command application. The list of commands is available with bardic -h:
usage: bardic [-h] SUBCOMMAND ...
Binomial RNA-DNA interaction caller.
optional arguments:
-h, --help show this help message and exit
Subcommands:
SUBCOMMAND
bed2h5 convert DNA parts to custom HDF5 format
binsizes Select bin size for each RNA and save it into dnah5 file
background
Create a bedGraph background track from DNA parts of selected RNAs
makerdc Create RDC file from dnah5 DNA parts and bedGraph background
track.
scaling Estimate scaling by fitting splines and adjust background
probabilities in RDC file.
peaks Estimate significance and fetch peaks at specified FDR level.
run Run pipeline with a single command.
simulate Simulate RNA-DNA data for a single RNA from first principles.
The command bardic run will run the whole pipeline, while other commands launch single steps of the pipeline. The command bardic simulate will simulate RNA-DNA contact data.
bardic run requires all input files as specified above. Additionally, algorithm can be tuned with a diverse set of parameters:
usage: bardic run [-h] [-f [{narrowPeak,bed}]] [-s [score_field]] [-mcon [int]]
[-tmin [int]] [-tmax [int]] [-tstep [int]] [-cmin [float]]
[-cmax [float]] [-cstep [float]] [-cstart [int]]
[-tol [float]] [-w [float]] [-bs [int]]
[-bt [{rnas,custom,uniform}]] [-i [float]] [-d [int]]
[-mt [float]] [-ns] [-nr] [-fv [numeric]] [-q [float]]
[-qt [{global,rna}]] [-c [int]]
dnaparts annotation chromsizes bgdata outdir
Run pipeline with a single command.
optional arguments:
-h, --help show this help message and exit
Input:
dnaparts BED6 file with coordinates of DNA parts. Names of
corresponding RNAs are in the "name" column.
annotation RNA annotation in BED format.
chromsizes If filename, then it is a UCSC headerless chromsizes
file; if genome abbreviation, then will fetch chromsizes
from UCSC
bgdata A file with data on background. If --bgtype="rnas", this
is a file with a list of RNAs with one RNA name per
line. If --bgtype="custom", this is a bedGraph file with
background signal in equally-sized bins. If
--bgtype="uniform", this is not used, write any string
here.
Output:
outdir Output directory name.
-f [{narrowPeak,bed}], --format [{narrowPeak,bed}]
Output peaks file format. (default: narrowPeak)
-s [score_field], --score [score_field]
If --format=bed, which value to fill the score field
with. If int, will fill every peak score with it; if
str, will take corresponding values from the column in
RDC (choices: bg_count, raw_bg_prob, scaling_factor,
bg_prob, signal_count, signal_prob, impute, fc, pvalue,
qvalue) (default: 0)
Binsize selection parameters:
-mcon [int], --min_contacts [int]
Minimal number of contacts to consider an RNA. Any RNA
with less contacts will be discarded from further
processing. (default: 1000)
-tmin [int], --trans_min [int]
Minimal trans bin size. (default: 10000)
-tmax [int], --trans_max [int]
Maximal trans bin size. (default: 1000000)
-tstep [int], --trans_step [int]
Step for increasing trans bin size. (default: 1000)
-cmin [float], --cis_min [float]
Minimal cis factor. (default: 1.1)
-cmax [float], --cis_max [float]
Maximal cis factor. (default: 2.0)
-cstep [float], --cis_step [float]
Step for inreasing cis factor. (default: 0.01)
-cstart [int], --cis_start [int]
Starting cis bin size. (default: 5000)
-tol [float], --tolerance [float]
Maximal absolute difference between two consecutive cost
function values to consider optimization converged.
(default: 0.01)
-w [float], --window [float]
Window size to average cost function values over.
(default: 1)
Background parameters:
-bs [int], --binsize [int]
Bin size of the background track. (default: 1000)
-bt [{rnas,custom,uniform}], --bgtype [{rnas,custom,uniform}]
Type of backround. If "rnas", then will calculate
background from trans-contacts of RNAs supplied as
"bgdata". If "custom", will use bedgraph track provided
as "bgdata". If "uniform", will use uniform background
with coverage 1. (default: rnas)
RDC creation parameters:
-i [float], --ifactor [float]
Imputation factor: if background coverage of a bin is 0,
this value is a multiplier of an average background
coverage to impute zero background coverage. (default:
0.01)
Scaling parameters:
-d [int], --degree [int]
Spline degree. (default: 3)
-mt [float], --max_threshold [float]
Maximal binomial test p-value to consider a point as an
outlier in a spline refinement procedure. (default:
0.05)
-ns, --no_scaling If included, do not estimate scaling. (default: False)
-nr, --no_refine If included, do not apply a spline refinement procedure.
(default: False)
-fv [numeric], --fill_value [numeric]
Fold-change fill ratio in case of 0/0. (default: 1)
Peaks parameters:
-q [float], --qval_threshold [float]
BH q-value threshold to consider bin a peak. (default:
0.05)
-qt [{global,rna}], --qval_type [{global,rna}]
BH q-value type to use for peak calling. If "global"
(default), will use q-values calculated for all RNAs; if
"rna", will use q-values calculated for each RNA
separately. (default: global)
Processing:
-c [int], --cores [int]
Maximal number of cores to use. (default: 1)
Arguments can be specified in the command line or in a separate config file that supplied like this:
bardic run @config.txt
The contents of @config.txt should look like this:
dnaparts
annotation
chromsizes
bgdata
outdir
-f
bed
-c
2
--no_scaling
In other words, the name of a CLI argument and its value are on different lines. Please note there is no line break at the end of the config file.
The output of bardic run will be put into a single directory. It consists of several files:
| File name | Description | File type |
|---|---|---|
| DnaDataset.dnah5 | All DNA parts of RNA-DNA contacts compactly stored with binning parameters for each RNA. Refer to schemas.md for the specification. | Custom HDF5 storage |
| background.bedGraph | A genomic track with background contacts. | bedGraph |
| contacts.rdc | Compact storage of binned contact profiles for each RNA along with the background track. Refer to schemas.md for the specification. | Custom HDF5 storage |
| selection.tsv | A table with statistics on bin size selection for each RNA. | Tab delimited file with header |
| peaks.narrowPeak or peaks.bed | Resulting peaks. | Tab-delimited file (narrowPeak or bed) |