view numeric_clustering.xml @ 27:6edcaa8dbb9f draft

"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ba6a47bdf76bbf4cb276206ac1a8cbf61332fd16"
author bgruening
date Fri, 13 Sep 2019 12:27:50 -0400
parents 37e193b3fdd7
children 83938131dd46
line wrap: on
line source

<tool id="sklearn_numeric_clustering" name="Numeric Clustering" version="@VERSION@">
    <description></description>
    <macros>
        <import>main_macros.xml</import>
    </macros>
    <expand macro="python_requirements"/>
    <expand macro="macro_stdio"/>
    <version_command>echo "@VERSION@"</version_command>
    <command><![CDATA[
    python "$cluster_script" '$inputs'
]]>
    </command>
    <configfiles>
        <inputs name="inputs"/>
        <configfile name="cluster_script">
<![CDATA[
import sys
import json
import numpy as np
import sklearn.cluster
import pandas

from sklearn import metrics
from scipy.io import mmread
from galaxy_ml.utils import read_columns


N_JOBS = int(__import__('os').environ.get('GALAXY_SLOTS', 1))

input_json_path = sys.argv[1]
with open(input_json_path, "r") as param_handler:
    params = json.load(param_handler)


selected_algorithm = params["input_types"]["algorithm_options"]["selected_algorithm"]

my_class = getattr(sklearn.cluster, selected_algorithm)
cluster_object = my_class()
options = params["input_types"]["algorithm_options"]["options"]

cluster_object.set_params(**options)
if 'n_jobs' in cluster_object.get_params():
    cluster_object.set_params( n_jobs=N_JOBS )

#if $input_types.selected_input_type == "sparse":
data_matrix = mmread("$infile")
#else:
data = pandas.read_csv("$infile", sep='\t', header=0, index_col=None, parse_dates=True, encoding=None, tupleize_cols=False )
header = 'infer' if params["input_types"]["header"] else None
column_option = params["input_types"]["column_selector_options"]["selected_column_selector_option"]
if column_option in ["by_index_number", "all_but_by_index_number", "by_header_name", "all_but_by_header_name"]:
    c = params["input_types"]["column_selector_options"]["col"]
else:
    c = None
data_matrix = read_columns(
    "$infile",
    c = c,
    c_option = column_option,
    sep='\t',
    header=header,
    parse_dates=True,
    encoding=None,
    tupleize_cols=False)
#end if

prediction = cluster_object.fit_predict( data_matrix )

if len(np.unique(prediction)) > 1:
    silhouette_score = metrics.silhouette_score(data_matrix,prediction,metric='euclidean')
else:
    silhouette_score = -1
sys.stdout.write('silhouette score:' + '\t' + str(silhouette_score) + '\n')

prediction_df = pandas.DataFrame(prediction, columns=["predicted"])

#if $input_types.selected_input_type == "sparse":
res = prediction_df
#else:
res = pandas.concat([data, prediction_df], axis=1)
#end if

res.to_csv(path_or_buf = "$outfile", sep="\t", index=False, header=False)
]]>
        </configfile>
    </configfiles>
    <inputs>
        <conditional name="input_types">
            <param name="selected_input_type" type="select" label="Select the format of input data">
                <option value="tabular" selected="true">Tabular Format (tabular, txt)</option>
                <option value="sparse">Sparse Vector Representation (mtx)</option>
            </param>
            <when value="sparse">
                <param name="infile" type="data" format="txt" label="Sparse vector (scipy.sparse.csr_matrix) file:" help="The following clustering algorithms support sparse matrix operations: ''Birch'', ''DBSCAN'', ''KMeans'', ''Mini BatchK Means'', and ''Spectral Clustering''. If your data is in tabular format, please use other clustering algorithms."/>
                <expand macro="clustering_algorithms_options"/>
            </when>
            <when value="tabular">
                <param name="infile" type="data" format="tabular" label="Data file with numeric values"/>
                <param name="header" type="boolean" optional="true" truevalue="booltrue" falsevalue="boolfalse" checked="True" label="Does the dataset contain header:" />
                <conditional name="column_selector_options">
                    <expand macro="samples_column_selector_options" col_name="col" multiple="true" infile="infile"/>
                </conditional>
                <!--expand macro="clustering_algorithms_options"-->
                <conditional name="algorithm_options">
                    <param name="selected_algorithm" type="select" label="Clustering Algorithm">
                        <option value="AgglomerativeClustering">Hierarchical Agglomerative Clustering</option>
                        <option value="AffinityPropagation">Affinity Propagation</option>
                        <option value="SpectralClustering">Spectral Clustering</option>
                        <option value="MiniBatchKMeans">Mini Batch KMeans</option>
                        <option value="MeanShift">MeanShift</option>
                        <option value="KMeans">KMeans</option>
                        <option value="DBSCAN">DBSCAN</option>
                        <option value="Birch">Birch</option>
                    </param>
                    <when value="KMeans">
                        <expand macro="kmeans_advanced_options"/>
                    </when>
                    <when value="DBSCAN">
                        <expand macro="dbscan_advanced_options"/>
                    </when>
                    <when value="Birch">
                        <expand macro="birch_advanced_options"/>
                    </when>
                    <when value="SpectralClustering">
                        <expand macro="spectral_clustering_advanced_options"/>
                    </when>
                    <when value="MiniBatchKMeans">
                        <expand macro="minibatch_kmeans_advanced_options"/>
                    </when>
                    <when value="AffinityPropagation">
                        <section name="options" title="Advanced Options" expanded="False">
                            <param argument="damping" type="float" optional="true" value="0.5" label="Damping factor" help="Damping factor between 0.5 and 1."/>
                            <expand macro="max_iter" default_value="200"/>
                            <param argument="convergence_iter" type="integer" optional="true" value="15" label="Number of iterations at each convergence step" help="Number of iterations with no change in the number of estimated clusters that stops the convergence."/>
                            <param argument="copy" type="boolean" optional="true" truevalue="booltrue" falsevalue="boolflase" checked="true" label="Copy" help="If False, the affinity matrix is modified inplace by the algorithm, for memory efficiency."/>
                            <!--param argument="preference"/-->
                            <param argument="affinity" type="select" label="Affinity" help="Affinity to use; euclidean uses the negative squared euclidean distance between points.">
                                <option value="euclidean">Euclidean</option>
                                <option value="precomputed">precomputed</option>
                            </param>
                        </section>
                    </when>
                    <when value="MeanShift">
                        <section name="options" title="Advanced Options" expanded="False">
                            <param argument="bandwidth" type="float" optional="true" value="" label="Kernel bandwidth" help="Bandwidth used in the RBF kernel. If not given, it will be computed using a heuristic based on the median of all pairwise distances."/>
                            <!--param argument="seeds"/-->
                            <param argument="bin_seeding" type="boolean" optional="true" truevalue="booltrue" falsevalue="boolflase" checked="true" label="Discretize initial kernel locations" help="If true, initial kernel locations are the bins grid whose coarseness corresponds to the bandwidth, speeding up the algorithm."/>
                            <param argument="min_bin_freq" type="integer" optional="true" value="1" label="Minimum number of seeds per bin" help="To speed up the algorithm, accept only those bins with at least min_bin_freq points as seeds."/>
                            <param argument="cluster_all" type="boolean" optional="true" truevalue="booltrue" falsevalue="boolflase" checked="true" label="Cluster all" help="If true, all points (including orphans) are clustered. If false, orphans are given cluster label -1."/>
                        </section>
                    </when>
                    <when value="AgglomerativeClustering">
                        <section name="options" title="Advanced Options" expanded="False">
                            <expand macro="n_clusters" default_value="2" />
                            <param argument="affinity" type="select" label="Affinity" help="Metric used to compute the linkage. If linkage is ''ward'', only ''euclidean'' is accepted.">
                                <option value="euclidean">Euclidean</option>
                                <option value="manhattan">Manhattan</option>
                                <option value="l1">L1</option>
                                <option value="l2">L2</option>
                                <option value="cosine">cosine</option>
                                <option value="precomputed">precomputed</option>
                            </param>
                            <!--param argument="memory"-->
                            <!--param argument="connectivity"-->
                            <!--param argument="n_components"/-->
                            <!--param argument="compute_full_tree"-->
                            <param argument="linkage" type="select" optional="true" label="Linkage" help="">
                                <option value="ward" selected="true">ward</option>
                                <option value="complete">complete</option>
                                <option value="average">average</option>
                                <option value="single">single</option>
                            </param>
                            <!--param argument="pooling_func"-->
                        </section>
                    </when>
                </conditional>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data format="tabular" name="outfile"/>
    </outputs>
    <tests>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="selected_algorithm" value="KMeans"/>
            <param name="col" value="2,3,4" />
            <param name="n_clusters" value="4" />
            <param name="init" value="k-means++" />
            <param name="random_state" value="100"/>
            <output name="outfile" file="cluster_result01.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="KMeans"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="n_clusters" value="4" />
            <param name="init" value="random" />
            <param name="random_state" value="100"/>
            <output name="outfile" file="cluster_result02.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="DBSCAN"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="algorithm" value="kd_tree"/>
            <param name="leaf_size" value="10"/>
            <param name="eps" value="1.0"/>
            <output name="outfile" file="cluster_result03.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="Birch"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="n_clusters" value="4"/>
            <param name="threshold" value="0.008"/>
            <output name="outfile" file="cluster_result04.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="Birch"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="branching_factor" value="20"/>
            <output name="outfile" file="cluster_result05.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="AffinityPropagation"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="affinity" value="euclidean"/>
            <param name="copy" value="false"/>
            <output name="outfile" file="cluster_result06.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="AffinityPropagation"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="damping" value="0.8"/>
            <output name="outfile" file="cluster_result07.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="MeanShift"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="min_bin_freq" value="3"/>
            <output name="outfile" file="cluster_result08.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="MeanShift"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="cluster_all" value="False"/>
            <output name="outfile" file="cluster_result09.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="AgglomerativeClustering"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="affinity" value="euclidean"/>
            <param name="linkage" value="average"/>
            <param name="n_clusters" value="4"/>
            <output name="outfile" file="cluster_result10.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="AgglomerativeClustering"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="linkage" value="complete"/>
            <param name="n_clusters" value="4"/>
            <output name="outfile" file="cluster_result11.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="SpectralClustering"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="eigen_solver" value="arpack"/>
            <param name="n_neighbors" value="12"/>
            <param name="n_clusters" value="4"/>
            <param name="assign_labels" value="discretize"/>
            <param name="random_state" value="100"/>
            <output name="outfile" file="cluster_result12.txt" compare="sim_size" />
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="SpectralClustering"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="assign_labels" value="discretize"/>
            <param name="random_state" value="100"/>
            <param name="degree" value="2"/>
            <output name="outfile" file="cluster_result13.txt" compare="sim_size" />
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="MiniBatchKMeans"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="2,3,4" />
            <param name="tol" value="0.5"/>
            <param name="random_state" value="100"/>
            <output name="outfile" file="cluster_result14.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="MiniBatchKMeans"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="n_init" value="5"/>
            <param name="col" value="2,3,4" />
            <param name="batch_size" value="10"/>
            <param name="n_clusters" value="4"/>
            <param name="random_state" value="100"/>
            <param name="reassignment_ratio" value="1.0"/>
            <output name="outfile" file="cluster_result15.txt"/>
        </test>
        <test>
            <param name="infile" value="numeric_values.tabular" ftype="tabular"/>
            <param name="selected_algorithm" value="KMeans"/>
            <param name="selected_input_type" value="tabular"/>
            <param name="col" value="1" />
            <param name="n_clusters" value="4" />
            <param name="random_state" value="100"/>
            <output name="outfile" file="cluster_result16.txt"/>
        </test>
        <test>
            <param name="infile" value="sparse.mtx" ftype="txt"/>
            <param name="selected_input_type" value="sparse"/>
            <param name="selected_algorithm" value="KMeans"/>
            <param name="n_clusters" value="2" />
            <param name="init" value="k-means++" />
            <param name="random_state" value="100"/>
            <output name="outfile" file="cluster_result17.txt"/>
        </test>
        <test>
            <param name="infile" value="sparse.mtx" ftype="txt"/>
            <param name="selected_algorithm" value="DBSCAN"/>
            <param name="selected_input_type" value="sparse"/>
            <param name="algorithm" value="kd_tree"/>
            <param name="leaf_size" value="10"/>
            <param name="eps" value="1.0"/>
            <output name="outfile" file="cluster_result18.txt"/>
        </test>
        <test>
            <param name="infile" value="sparse.mtx" ftype="txt"/>
            <param name="selected_algorithm" value="Birch"/>
            <param name="selected_input_type" value="sparse"/>
            <param name="n_clusters" value="2"/>
            <param name="threshold" value="0.008"/>
            <output name="outfile" file="cluster_result19.txt"/>
        </test>
        <test>
            <param name="infile" value="sparse.mtx" ftype="txt"/>
            <param name="selected_algorithm" value="MiniBatchKMeans"/>
            <param name="selected_input_type" value="sparse"/>
            <param name="n_init" value="5"/>
            <param name="batch_size" value="10"/>
            <param name="n_clusters" value="2"/>
            <param name="random_state" value="100"/>
            <param name="reassignment_ratio" value="1.0"/>
            <output name="outfile" file="cluster_result20.txt"/>
        </test>
        <test>
            <param name="infile" value="sparse.mtx" ftype="txt"/>
            <param name="selected_algorithm" value="SpectralClustering"/>
            <param name="selected_input_type" value="sparse"/>
            <param name="assign_labels" value="discretize"/>
            <param name="n_clusters" value="2"/>
            <param name="random_state" value="100"/>
            <param name="degree" value="2"/>
            <output name="outfile" file="cluster_result21.txt"/>
        </test>
    </tests>
    <help><![CDATA[
**What it does**
This tool offers different clustering algorithms which are provided by
scikit-learn to find similarities among samples and cluster the samples based on these similarities.
    ]]></help>
    <expand macro="sklearn_citation"/>
</tool>