% Functional Partitioning % J. Izaak Miller

Functional Partitioning: A Software Package for RWR-based Gene Cluster Identification

• Integrate multi-omics data using multiplex networks.
• Identify functionally-related groups of genes using random walk with restart on multiplex networks.

Installation

Standard installation:

$ git clone https://github.com/izaakm/jail-functional-partitioning
$ cd jail-functional-partitioning
$ make

If you prefer to install dependencies with conda, you can use the provided environment.yml file:

$ git clone https://github.com/izaakm/jail-functional-partitioning
$ cd jail-functional-partitioning
$ conda env update -f ./environment.yml
$ make

Or create a new environment for 'functional-partitioning':

$ git clone https://github.com/izaakm/jail-functional-partitioning
$ cd jail-functional-partitioning
$ conda env create -f ./environment.yml
$ conda activate functional-partitioning
$ make

Verify the installation:

$ functional_partitioning --help

Usage

usage: functional_partitioning [-h] [--rwr-fullranks RWR_FULLRANKS]
                               [--nodetable NODETABLE] [--partition]
                               [--no-partition] [--cut-threshold CUT_THRESHOLD]
                               [--cut-method {dynamic,hard,none}]
                               [--dendrogram-style
                               {rectangular,r,polar,p,none,n}]
                               [--labels-use-clusters] [--labels-use-names
                               [LABELS_USE_NAMES ...]] [--labels-use-locs
                               [LABELS_USE_LOCS ...]] [--labels-sep LABELS_SEP]
                               [--outdir OUTDIR] [--out-dendrogram
                               OUT_DENDROGRAM] [--no-plot] [--out-clusters
                               OUT_CLUSTERS] [--no-clusters]
                               [--path-to-conda-env PATH_TO_CONDA_ENV]
                               [--path-to-rwrtoolkit PATH_TO_RWRTOOLKIT]
                               [--multiplex MULTIPLEX] [--geneset GENESET]
                               [--method METHOD] [--folds FOLDS] [--restart
                               RESTART] [--tau TAU] [--numranked NUMRANKED]
                               [--modname MODNAME] [--plot PLOT] [--threads
                               THREADS] [--init-test-fullranks
                               INIT_TEST_FULLRANKS] [--verbose] [--version]

Partition seeds from `RWR-CV --method=singletons ...` into clusters.

optional arguments:
  -h, --help            show this help message and exit
  --rwr-fullranks RWR_FULLRANKS, -f RWR_FULLRANKS
                        Path to "fullranks" file from `RWR-CV
                        --method=singletons ...` (default: None)
  --nodetable NODETABLE
                        Path to "nodetable" file. This is a TSV file where the
                        first column is the node name (i.e., the seed genes
                        from RWR-fullranks). (default: None)
  --partition, -p       Perform functional partitioning on "seed genes" from
                        RWR fullranks file. This is the default. (default: True)
  --no-partition        Do not perform functional partitioning. (default: True)
  --cut-threshold CUT_THRESHOLD, -t CUT_THRESHOLD
                        Cut the dendrogram at this threshold. Only used if
                        `--cut-method=hard`. (default: 0.3)
  --cut-method {dynamic,hard,none}, -m {dynamic,hard,none}
                        If `dynamic`, use dynamicTreeCut to determine clusters
                        without a hard threshold. If `none`, return all the
                        clusterings. Otherwise, use with `--cut-threshold` to
                        provide a numeric cut threshold. (default: dynamic)
  --dendrogram-style {rectangular,r,polar,p,none,n}, -s {rectangular,r,polar,p,none,n}
                        Plot the dendrogram in rectangular or polar
                        coordinates. If "none", then do not plot the dendrogram
                        (this is redundant with --no-plot). (default:
                        rectangular)
  --labels-use-clusters
  --labels-use-names [LABELS_USE_NAMES ...]
                        Label the dendrogram using columns from the nodetable.
                        This is a space-separated list of column names from the
                        nodetable. Pass columns as strings (column names).
                        (default: None)
  --labels-use-locs [LABELS_USE_LOCS ...]
                        Label the dendrogram using columns from the nodetable.
                        This is a space-separated list of integers indicating
                        columns from the nodetable (0-index, e.g., the first
                        column, which contains the node names, has index 0;
                        the second column has index 1, etc). (default: None)
  --labels-sep LABELS_SEP
                        The separator that will be used if multiple columns
                        from nodetable are used to label the dendrogram.
                        (default: | )
  --outdir OUTDIR       Save dendrogram and clusters to path. (default: None)
  --out-dendrogram OUT_DENDROGRAM, -d OUT_DENDROGRAM
                        Save dendrogram to path. (default: None)
  --no-plot             Do not plot the dendrogram. (default: False)
  --out-clusters OUT_CLUSTERS, -c OUT_CLUSTERS
                        Save clusters to path as tsv file with columns
                        "label", "cluster". When --threshold is 0 (the
                        default) each gene is put into a separate cluster
                        (i.e., every cluster has only a single gene).
                        (default: None)
  --no-clusters         Do not export clusters to file. (default: False)
  --path-to-conda-env PATH_TO_CONDA_ENV
  --path-to-rwrtoolkit PATH_TO_RWRTOOLKIT
  --multiplex MULTIPLEX
  --geneset GENESET
  --method METHOD
  --folds FOLDS
  --restart RESTART
  --tau TAU
  --numranked NUMRANKED
  --modname MODNAME
  --plot PLOT
  --threads THREADS
  --init-test-fullranks INIT_TEST_FULLRANKS
                        Create fullranks file for testing at the given path. (default: None)
  --verbose, -v         Default: WARNING; once: INFO; twice: DEBUG (default: 0)
  --version             Print version and exit. (default: False)

Examples

If you have a gene set and a multiplex network, run RWR and partition the geneset:

functional_partitioning \
    --path-to-rwrtoolkit <path to RWRtoolkit directory> \
    --multiplex <path to multiplex object> \
    --geneset <path to your gene set> \
    --outdir <path to output directory>

If you already have results from RWRtoolkit, you can pass in your 'fullranks' file:

functional_partitioning \
    --rwr-fullranks <path to fullranks file> \
    --outdir <path to output directory>

The default cut method is to use dynamicTreeCut, explicitly:

functional_partitioning \
    --rwr-fullranks <path to fullranks file> \
    --cut-method dynamic \
    --outdir <path to output directory>

You can change this to use a hard threshold instead:

functional_partitioning \
    --rwr-fullranks <path to fullranks file> \
    --cut-method hard \
    --cut-threshold 0.3 \
    --outdir <path to output directory>

You can change the labels on the leaves of the dendrogram. This requires that you have a nodetable file. The nodetable is a TSV file with the first column as the seed genes use for RWR-CV (singletons), e.g., something like this:

gene_id          gene_info
Potri.001G377800 This is gene 001G377800.
Potri.001G429430 This is gene 001G429430.
Potri.003G099600 This is gene 003G099600.
...

Then, provide the --nodetable and --labels-use-{locs,names} parameters to tell this script which columns to use as labels on the dendrogram. E.g., to use the column "gene_info" as the labels, you can either provide the column by name:

functional_partitioning \
    --rwr-fullranks <path to fullranks file> \
    --nodetable <path to nodetable.tsv file> \
    --labels-use-names "gene_info"

Or provide the column by number:

functional_partitioning \
    --rwr-fullranks <path to fullranks file> \
    --nodetable <path to nodetable.tsv file> \
    --labels-use-locs 1

If you want to keep the node ID in the label, provide that column as one of the labels in addition to any others:

functional_partitioning \
    --rwr-fullranks <path to fullranks file> \
    --nodetable <path to nodetable.tsv file> \
    --labels-use-names "gene_id" "gene_info"

Acknowledgements

Derived from the cookiecutter data science project template.