Usage

Contents:

• Input data
  • Mandatory
  • Optional
• Running RCK
  • preprocessing options
  • running options
• Examples

Input data

Mandatory

RCK expects two mandatory peaces of the input:

• --scnt - clone- and allele-specific segment copy number predictions, obtained from 3rd-party tools (see segments docs for more details).
• --adjacencies (unlabeled) novel adjacencies (aka SVs) obtained from 3rd-party tools (see adjacencies docs for more details).

Both inputs must be i the RCK format (refer to segments and adjacencies docs on the formatting issues.)

Optional

• --adjacency-groups - Adjacencies groups (see adjacencies groups docs for more details).
• --clone-ids - a comma-separated list of clone ids (as present in the scnt input).
• advanced --telomere-positions and --telomere-segments - Telomeres (either via exact locations, or via segments, for which all spanned extremities will be considered as possible additional telomeres)
• advanced --fragments - Fragments that span segments. Only works properly if no preprocessing on input scnt is performed.

Running RCK

Running RCK inference algorithm is achieved through the rck executable (which is automatically added to your PATH with RCK installation).

preprocessing options

When running RCK, a lot of input preprocessing options, that are achieved via RCK utilities.

All preprocessing can be turned off via the --no-pre flag, but this is an advanced option, use with caution.

All the --pre-scnt-xxx flags refer to --scnt input clone- and allele-specific segment copy number values (in RCK format) preprocessing, that is similar to the refine command on the rck-scnt-process tool.

All the --pre-scnb-xxx flags refer to creating copy number boundaries for the inferred copy number values. By default the strategy for obtaining the copy number boundaries is the uniform min-max one, where regardless of the input clone- and allele-specific copy number, the lower bound is set to the --pre-scnb-uniform-min value (default 0) and the upper is set to the --pre-scnb-uniform-max value (default 10). When working with genomes with known highly amplified segments, one can think about altering the default value for the --pre-scnb-uniform-max.

Preprocessing of the input adjacencies concerns the reciprocality adjustments, and is similar to the rck-adj-process reciprocal command and the --pre-adj-xxx option mirror those of the rck-adj-process reciprocal. Adjacency preprocessing can be turned off by specifying the --pre-no-adj.

running options

One of the main arguments in running rck is the --workdir option, specifying the working directory in which the three following directories are created:

• raw_input - contains exact copies of the input files
• input - contains fully preprocessed data
• output - contains inference results from the RCK algorithm

Running options for rck start with the --run- prefix. Running RCK without actually executing the inference algorithm can be achieved by using the --no-run flag. This will prevent the actual gurobi based ilp solving and respective karyotype inference, but will preprocess (unless disabled) of all the input, and putting the preprocessed data into the workdir/input directory.

The --run-g- are the flags corresponding to setting Gurobi related options:

• --run-g-mip-gap - the gap between the best bound and best objective, after which the Gurobi solver will stop crunching numbers (default: 0.015, or 1.5% difference)
• --run-g-time-limit - the maximum time (in seconds) for gurobi to run, before stopping execution and taking the current best objective as the result (default: 28800, aka 8 hours)
• --run-g-threads - number of threads gurobi will use (deault: 4)
• --run-g-allow-interrupted - allow for gurobi run to be interrupted and still use the best obtained objective for the inference result

Other flags:

• --run-nas-fp - default False Positive upper bound (i.e., at most a --run-nas-fp fraction of input novel adjacencies can be not used in the inferred karyotypes). Default is 0.1
• --run-group-m-default-fp - default False Positive values for molecule adjacencies groups (unless explicitly specified in with the fp value in the extra field). Default is 0.1
• --run-segment-length-attr - an choice based attribute that is used to get the segments length. Default is length_100 which means that for every segment of length l an ceil(l/100) value is used in the inference minimization.

Examples

The following command runs RCK inference on the clone- and allele-specific segment copy number tensor (stored in the input.rck.scnt.tsv)