|
| 1 | +{ |
| 2 | + "cells": [ |
| 3 | + { |
| 4 | + "cell_type": "markdown", |
| 5 | + "metadata": {}, |
| 6 | + "source": [ |
| 7 | + "<img src=\"../static/images/joinnode.png\" width=\"240\">\n", |
| 8 | + "\n", |
| 9 | + "# JoinNode\n", |
| 10 | + "\n", |
| 11 | + "JoinNode have the opposite effect of a [MapNode](basic_mapnodes.ipynb) or [iterables](basic_iteration.ipynb). Where they split up the execution workflow into many different branches, a JoinNode merges them back into on node. For a more detailed explanation, check out [JoinNode, synchronize and itersource](http://nipype.readthedocs.io/en/latest/users/joinnode_and_itersource.html) from the main homepage." |
| 12 | + ] |
| 13 | + }, |
| 14 | + { |
| 15 | + "cell_type": "markdown", |
| 16 | + "metadata": {}, |
| 17 | + "source": [ |
| 18 | + "## Simple example\n", |
| 19 | + "\n", |
| 20 | + "Let's consider the very simple example depicted at the top of this page:" |
| 21 | + ] |
| 22 | + }, |
| 23 | + { |
| 24 | + "cell_type": "code", |
| 25 | + "execution_count": null, |
| 26 | + "metadata": { |
| 27 | + "collapsed": false |
| 28 | + }, |
| 29 | + "outputs": [], |
| 30 | + "source": [ |
| 31 | + "from nipype import Node, JoinNode, Workflow\n", |
| 32 | + "\n", |
| 33 | + "# Specify fake input node A\n", |
| 34 | + "a = Node(interface=A(), name=\"a\")\n", |
| 35 | + "\n", |
| 36 | + "# Iterate over fake node B's input 'in_file?\n", |
| 37 | + "b = Node(interface=B(), name=\"b\")\n", |
| 38 | + "b.iterables = ('in_file', [file1, file2])\n", |
| 39 | + "\n", |
| 40 | + "# Pass results on to fake node C\n", |
| 41 | + "c = Node(interface=C(), name=\"c\")\n", |
| 42 | + "\n", |
| 43 | + "# Join forked execution workflow in fake node D\n", |
| 44 | + "d = JoinNode(interface=D(),\n", |
| 45 | + " joinsource=\"b\",\n", |
| 46 | + " joinfield=\"in_files\",\n", |
| 47 | + " name=\"d\")\n", |
| 48 | + "\n", |
| 49 | + "# Put everything into a workflow as usual\n", |
| 50 | + "workflow = Workflow(name=\"workflow\")\n", |
| 51 | + "workflow.connect([(a, b, [('subject', 'subject')]),\n", |
| 52 | + " (b, c, [('out_file', 'in_file')])\n", |
| 53 | + " (c, d, [('out_file', 'in_files')])\n", |
| 54 | + " ])" |
| 55 | + ] |
| 56 | + }, |
| 57 | + { |
| 58 | + "cell_type": "markdown", |
| 59 | + "metadata": {}, |
| 60 | + "source": [ |
| 61 | + "As you can see, setting up a ``JoinNode`` is rather simple. The only difference to a normal ``Node`` are the ``joinsource`` and the ``joinfield``. ``joinsource`` specifies from which node the information to join is coming and the ``joinfield`` specifies the input field of the JoinNode where the information to join will be entering the node." |
| 62 | + ] |
| 63 | + }, |
| 64 | + { |
| 65 | + "cell_type": "markdown", |
| 66 | + "metadata": {}, |
| 67 | + "source": [ |
| 68 | + "## More realistic example\n", |
| 69 | + "\n", |
| 70 | + "Let's consider another example where we have one node that iterates over 3 different numbers and another node that joins those three different numbers (each coming from a separate branch of the workflow) into one list. To make the whole thing a bit more realistic, the second node will use the ``Function`` interface to do something with those numbers, before we spit them out again." |
| 71 | + ] |
| 72 | + }, |
| 73 | + { |
| 74 | + "cell_type": "code", |
| 75 | + "execution_count": null, |
| 76 | + "metadata": { |
| 77 | + "collapsed": true |
| 78 | + }, |
| 79 | + "outputs": [], |
| 80 | + "source": [ |
| 81 | + "from nipype import JoinNode, Node, Workflow\n", |
| 82 | + "from nipype.interfaces.utility import Function, IdentityInterface" |
| 83 | + ] |
| 84 | + }, |
| 85 | + { |
| 86 | + "cell_type": "code", |
| 87 | + "execution_count": null, |
| 88 | + "metadata": { |
| 89 | + "collapsed": true |
| 90 | + }, |
| 91 | + "outputs": [], |
| 92 | + "source": [ |
| 93 | + "# Create iteration node\n", |
| 94 | + "from nipype import IdentityInterface\n", |
| 95 | + "iternode = Node(IdentityInterface(fields=['number_id']),\n", |
| 96 | + " name=\"iternode\")\n", |
| 97 | + "iternode.iterables = [('number_id', [1, 4, 9])]" |
| 98 | + ] |
| 99 | + }, |
| 100 | + { |
| 101 | + "cell_type": "code", |
| 102 | + "execution_count": null, |
| 103 | + "metadata": { |
| 104 | + "collapsed": false |
| 105 | + }, |
| 106 | + "outputs": [], |
| 107 | + "source": [ |
| 108 | + "# Create join node - compute square root for each element in the joined list\n", |
| 109 | + "def compute_sqrt(numbers):\n", |
| 110 | + " from math import sqrt\n", |
| 111 | + " return [sqrt(e) for e in numbers]\n", |
| 112 | + "\n", |
| 113 | + "joinnode = JoinNode(Function(input_names=['numbers'],\n", |
| 114 | + " output_names=['sqrts'],\n", |
| 115 | + " function=compute_sqrt),\n", |
| 116 | + " name='joinnode',\n", |
| 117 | + " joinsource='iternode',\n", |
| 118 | + " joinfield=['numbers'])" |
| 119 | + ] |
| 120 | + }, |
| 121 | + { |
| 122 | + "cell_type": "code", |
| 123 | + "execution_count": null, |
| 124 | + "metadata": { |
| 125 | + "collapsed": false |
| 126 | + }, |
| 127 | + "outputs": [ |
| 128 | + { |
| 129 | + "name": "stdout", |
| 130 | + "output_type": "stream", |
| 131 | + "text": [ |
| 132 | + "170306-22:38:22,861 workflow INFO:\n", |
| 133 | + "\t Workflow joinflow settings: ['check', 'execution', 'logging']\n", |
| 134 | + "170306-22:38:22,871 workflow INFO:\n", |
| 135 | + "\t Running serially.\n", |
| 136 | + "170306-22:38:22,873 workflow INFO:\n", |
| 137 | + "\t Executing node joinnode in dir: /tmp/tmpm8NCMb/joinflow/joinnode\n" |
| 138 | + ] |
| 139 | + } |
| 140 | + ], |
| 141 | + "source": [ |
| 142 | + "# Create the workflow and run it\n", |
| 143 | + "joinflow = Workflow(name='joinflow')\n", |
| 144 | + "joinflow.connect(iternode, 'number_id', joinnode, 'numbers')\n", |
| 145 | + "res = joinflow.run()" |
| 146 | + ] |
| 147 | + }, |
| 148 | + { |
| 149 | + "cell_type": "markdown", |
| 150 | + "metadata": {}, |
| 151 | + "source": [ |
| 152 | + "Now, let's look at the input and output of the joinnode:" |
| 153 | + ] |
| 154 | + }, |
| 155 | + { |
| 156 | + "cell_type": "code", |
| 157 | + "execution_count": null, |
| 158 | + "metadata": { |
| 159 | + "collapsed": false |
| 160 | + }, |
| 161 | + "outputs": [ |
| 162 | + { |
| 163 | + "data": { |
| 164 | + "text/plain": [ |
| 165 | + "\n", |
| 166 | + "sqrts = [1.0, 2.0, 3.0]" |
| 167 | + ] |
| 168 | + }, |
| 169 | + "execution_count": null, |
| 170 | + "metadata": {}, |
| 171 | + "output_type": "execute_result" |
| 172 | + } |
| 173 | + ], |
| 174 | + "source": [ |
| 175 | + "res.nodes()[0].result.outputs" |
| 176 | + ] |
| 177 | + }, |
| 178 | + { |
| 179 | + "cell_type": "code", |
| 180 | + "execution_count": null, |
| 181 | + "metadata": { |
| 182 | + "collapsed": false |
| 183 | + }, |
| 184 | + "outputs": [ |
| 185 | + { |
| 186 | + "data": { |
| 187 | + "text/plain": [ |
| 188 | + "\n", |
| 189 | + "function_str = <undefined>\n", |
| 190 | + "ignore_exception = <undefined>\n", |
| 191 | + "numbers = <undefined>\n", |
| 192 | + "numbersJ1 = 1\n", |
| 193 | + "numbersJ2 = 4\n", |
| 194 | + "numbersJ3 = 9" |
| 195 | + ] |
| 196 | + }, |
| 197 | + "execution_count": null, |
| 198 | + "metadata": {}, |
| 199 | + "output_type": "execute_result" |
| 200 | + } |
| 201 | + ], |
| 202 | + "source": [ |
| 203 | + "res.nodes()[0].inputs" |
| 204 | + ] |
| 205 | + } |
| 206 | + ], |
| 207 | + "metadata": { |
| 208 | + "anaconda-cloud": {}, |
| 209 | + "kernelspec": { |
| 210 | + "display_name": "Python [conda root]", |
| 211 | + "language": "python", |
| 212 | + "name": "conda-root-py" |
| 213 | + }, |
| 214 | + "language_info": { |
| 215 | + "codemirror_mode": { |
| 216 | + "name": "ipython", |
| 217 | + "version": 2 |
| 218 | + }, |
| 219 | + "file_extension": ".py", |
| 220 | + "mimetype": "text/x-python", |
| 221 | + "name": "python", |
| 222 | + "nbconvert_exporter": "python", |
| 223 | + "pygments_lexer": "ipython2", |
| 224 | + "version": "2.7.13" |
| 225 | + } |
| 226 | + }, |
| 227 | + "nbformat": 4, |
| 228 | + "nbformat_minor": 0 |
| 229 | +} |
0 commit comments