02 - Figure SF4.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "# %matplotlib inline\n",
    "import os\n",
    "from pathlib import Path\n",
    "import warnings\n",
    "\n",
    "import numpy as np\n",
    "import nibabel as nb\n",
    "import pandas as pd\n",
    "\n",
    "import matplotlib as mpl\n",
    "mpl.use('pgf')\n",
    "from matplotlib import pyplot as plt\n",
    "from matplotlib import gridspec\n",
    "import seaborn as sn\n",
    "import palettable\n",
    "\n",
    "from nilearn.image import concat_imgs, mean_img\n",
    "from nilearn import plotting\n",
    "\n",
    "warnings.simplefilter('ignore')\n",
    "\n",
    "DATA_HOME = Path(os.getenv('FMRIPREP_DATA_HOME', os.getcwd())).resolve()\n",
    "DS030_HOME = DATA_HOME / 'ds000030' / '1.0.3'\n",
    "DERIVS_HOME = DS030_HOME / 'derivatives'\n",
    "ATLAS_HOME = DATA_HOME / 'templates' / 'mni_icbm152_nlin_asym_09c'\n",
    "ANALYSIS_HOME = DERIVS_HOME / 'fmriprep_vs_feat_2.0-oe'\n",
    "\n",
    "fprep_home = DERIVS_HOME / 'fmriprep_1.0.8' / 'fmriprep'\n",
    "feat_home = DERIVS_HOME / 'fslfeat_5.0.10' / 'featbids'\n",
    "\n",
    "out_folder = Path(os.getenv('FMRIPREP_OUTPUTS') or '').resolve()\n",
    "\n",
    "# Load MNI152 nonlinear, asymmetric 2009c atlas\n",
    "atlas = nb.load(str(ATLAS_HOME / '1mm_T1.nii.gz'))\n",
    "mask1mm = nb.load(str(ATLAS_HOME / '1mm_brainmask.nii.gz')).get_data() > 0\n",
    "mask2mm = nb.load(str(ATLAS_HOME / '2mm_brainmask.nii.gz')).get_data() > 0\n",
    "data = atlas.get_data()\n",
    "data[~mask1mm] = data[~mask1mm].max()\n",
    "atlas = nb.Nifti1Image(data, atlas.affine, atlas.header)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pgf_with_custom_preamble = {\n",
    "    'text.usetex': True,    # use inline math for ticks\n",
    "    'pgf.rcfonts': False,   # don't setup fonts from rc parameters\n",
    "    'pgf.texsystem': 'xelatex',\n",
    "    'verbose.level': 'debug-annoying',\n",
    "    \"pgf.preamble\": [\n",
    "        r\"\"\"\\usepackage{fontspec}\n",
    "\\setsansfont{HelveticaLTStd-Light}[\n",
    "Extension=.otf,\n",
    "BoldFont=HelveticaLTStd-Bold,\n",
    "ItalicFont=HelveticaLTStd-LightObl,\n",
    "BoldItalicFont=HelveticaLTStd-BoldObl,\n",
    "]\n",
    "\\setmainfont{HelveticaLTStd-Light}[\n",
    "Extension=.otf,\n",
    "BoldFont=HelveticaLTStd-Bold,\n",
    "ItalicFont=HelveticaLTStd-LightObl,\n",
    "BoldItalicFont=HelveticaLTStd-BoldObl,\n",
    "]\n",
    "\\setmonofont{Inconsolata-dz}\n",
    "\"\"\",\n",
    "        r'\\renewcommand\\familydefault{\\sfdefault}',\n",
    "#         r'\\setsansfont[Extension=.otf]{Helvetica-LightOblique}',\n",
    "#         r'\\setmainfont[Extension=.ttf]{DejaVuSansCondensed}',\n",
    "#         r'\\setmainfont[Extension=.otf]{FiraSans-Light}',\n",
    "#         r'\\setsansfont[Extension=.otf]{FiraSans-Light}',\n",
    "    ]\n",
    "}\n",
    "mpl.rcParams.update(pgf_with_custom_preamble)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def mean_std_map(pipe_home, meanmask, force=False, lazy=False, maskval=1000):\n",
    "    pipe_std = pipe_home / 'summary_stdev.nii.gz'\n",
    "    pipe_mean = pipe_home / 'summary_means.nii.gz'\n",
    "\n",
    "    if force or not pipe_mean.is_file():\n",
    "        all_mus = []\n",
    "        if lazy:\n",
    "            all_mus = [nb.load(str(f)) for f in pipe_home.glob(\n",
    "                'sub-*/func/sub-*_task-stopsignal_bold_space-MNI152NLin2009cAsym_avgpreproc.nii.gz')]\n",
    "        \n",
    "        if not all_mus:\n",
    "            print('Generating means file')\n",
    "            pipe_files = list(pipe_home.glob(\n",
    "                'sub-*/func/sub-*_task-stopsignal_bold_space-MNI152NLin2009cAsym_preproc.nii.gz'))\n",
    "            all_mus = []\n",
    "            for f in pipe_files:\n",
    "                mean = mean_img(str(f))\n",
    "                data = mean.get_data()\n",
    "                sigma = np.percentile(data[meanmask], 50) / maskval\n",
    "                data /= sigma\n",
    "                all_mus.append(nb.Nifti1Image(data, mean.affine, mean.header))\n",
    "                \n",
    "        meannii = concat_imgs(all_mus, auto_resample=False)\n",
    "        meannii.to_filename(str(pipe_mean))\n",
    "        force = True\n",
    "\n",
    "    if force or not pipe_std.is_file():\n",
    "        print('Generating standard deviation map')\n",
    "        meannii = nb.load(str(pipe_mean))\n",
    "        nb.Nifti1Image(meannii.get_data().std(3), meannii.affine, meannii.header).to_filename(str(pipe_std))\n",
    "        \n",
    "    return pipe_mean, pipe_std\n",
    "\n",
    "# Use the WM mask to normalize intensities of EPI means\n",
    "meanmask = nb.load(str(ATLAS_HOME / '2mm_tpm_wm.nii.gz')).get_data() > 0.9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Calculate average and std\n",
    "fprep_mean, fprep_std = mean_std_map(fprep_home, meanmask)\n",
    "feat_mean, feat_std = mean_std_map(feat_home, meanmask)\n",
    "\n",
    "# Trick to avoid nilearn zooming in\n",
    "feat_nii = nb.load(str(feat_std))\n",
    "\n",
    "fd = feat_nii.get_data()\n",
    "fd[0, 0, :] = 50\n",
    "fd[0, -1, :] = 50\n",
    "fd[-1, 0, :] = 50\n",
    "fd[-1, -1, :] = 50\n",
    "\n",
    "nb.Nifti1Image(fd, feat_nii.affine, feat_nii.header).to_filename('newfeat.nii.gz')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.0, 0.009324209800697824, -13.084191226959231, 571.25822343826292)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "plt.clf()\n",
    "fig = plt.gcf()\n",
    "_ = fig.set_size_inches(15, 7)\n",
    "gs = gridspec.GridSpec(2, 2, width_ratios=[10, 1], height_ratios=[1, 1],  hspace=0.0, wspace=0.03)\n",
    "\n",
    "b_ax1 = plt.subplot(gs[0, 0])\n",
    "b_ax2 = plt.subplot(gs[1, 0])\n",
    "\n",
    "thres = 20\n",
    "vmin = 50\n",
    "vmax = 200\n",
    "\n",
    "disp = plotting.plot_anat(str(fprep_std), display_mode='z', annotate=False,\n",
    "                          cut_coords=[-15, -5, 10, 20, 40], cmap='cividis', threshold=thres, vmin=vmin, vmax=vmax,\n",
    "                          axes=b_ax1, resampling_interpolation='nearest')\n",
    "disp.annotate(size=12, left_right=True, positions=True)\n",
    "disp.add_contours(str(ATLAS_HOME / '1mm_tpm_csf.nii.gz'), colors=['k'], levels=[0.8])\n",
    "disp.add_contours(str(ATLAS_HOME / '1mm_tpm_wm.nii.gz'), colors=['w'], levels=[0.8], linewidths=[1], alpha=0.7)\n",
    "disp.add_contours(str(ATLAS_HOME / '1mm_brainmask.nii.gz'), colors=['k'], levels=[0.8], linewidths=[3], alpha=.7)\n",
    "\n",
    "\n",
    "disp = plotting.plot_anat('newfeat.nii.gz', display_mode='z', annotate=False,\n",
    "                          cut_coords=[-15, -5, 10, 20, 40], cmap='cividis', threshold=thres, vmin=vmin, vmax=vmax,\n",
    "                          axes=b_ax2, resampling_interpolation='nearest')\n",
    "disp.annotate(size=12, left_right=False, positions=False)\n",
    "disp.annotate(size=12, left_right=False, positions=False, scalebar=True,\n",
    "              loc=3, size_vertical=2, label_top=False, frameon=True, borderpad=0.1)\n",
    "\n",
    "disp.add_contours(str(ATLAS_HOME / '1mm_tpm_csf.nii.gz'), colors=['k'], levels=[0.8])\n",
    "disp.add_contours(str(ATLAS_HOME / '1mm_tpm_wm.nii.gz'), colors=['w'], levels=[0.8], linewidths=[1], alpha=0.7)\n",
    "disp.add_contours(str(ATLAS_HOME / '1mm_brainmask.nii.gz'), colors=['k'], levels=[0.8], linewidths=[3], alpha=.7)\n",
    "\n",
    "b_ax1.annotate(\n",
    "    'fMRIPrep',\n",
    "    xy=(0., .5), xycoords='axes fraction', xytext=(-20, .0),\n",
    "    textcoords='offset points', va='center', color='k', size=15,\n",
    "    rotation=90)\n",
    "\n",
    "b_ax2.annotate(\n",
    "    r'\\texttt{feat}',\n",
    "    xy=(0., .5), xycoords='axes fraction', xytext=(-20, .0),\n",
    "    textcoords='offset points', va='center', color='k', size=12,\n",
    "    rotation=90)\n",
    "\n",
    "\n",
    "inner_grid = gridspec.GridSpecFromSubplotSpec(1, 2, width_ratios=[1.5, 15],\n",
    "                                              subplot_spec=gs[:, -1], wspace=0.01)\n",
    "\n",
    "b_ax3 = fig.add_subplot(inner_grid[0])\n",
    "gradient = np.hstack((np.zeros((50,)), np.linspace(0, 1, 120), np.ones((130,))))[::-1]\n",
    "gradient = np.vstack((gradient, gradient))\n",
    "b_ax3.imshow(gradient.T, aspect='auto', cmap=plt.get_cmap('cividis'))\n",
    "b_ax3.xaxis.set_ticklabels([])\n",
    "b_ax3.xaxis.set_ticks([])\n",
    "b_ax3.yaxis.set_ticklabels([])\n",
    "b_ax3.yaxis.set_ticks([])\n",
    "\n",
    "b_ax4 = fig.add_subplot(inner_grid[1])\n",
    "sn.distplot(nb.load(str(fprep_std)).get_data()[mask2mm], label='fMRIPrep',\n",
    "            vertical=True, ax=b_ax4, kde=False, norm_hist=True)\n",
    "sn.distplot(nb.load(str(feat_std)).get_data()[mask2mm], label=r'\\texttt{feat}', vertical=True,\n",
    "            color='darkorange', ax=b_ax4, kde=False, norm_hist=True)\n",
    "\n",
    "# plt.gca().set_ylim((0, 300))\n",
    "plt.legend(prop={'size': 15}, edgecolor='none')\n",
    "\n",
    "b_ax4.xaxis.set_ticklabels([])\n",
    "b_ax4.xaxis.set_ticks([])\n",
    "b_ax4.yaxis.set_ticklabels([])\n",
    "b_ax4.yaxis.set_ticks([])\n",
    "\n",
    "plt.axis('off')\n",
    "b_ax3.axis('off')\n",
    "b_ax4.axis('off')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "feat_closeup = nb.load(str(feat_std))\n",
    "feat_data = feat_closeup.get_data()[15:82, 72:106, :]\n",
    "nb.Nifti1Image(feat_data, feat_closeup.affine, feat_closeup.header).to_filename('feat_clip.nii.gz')\n",
    "\n",
    "fprep_closeup = nb.load(str(fprep_std))\n",
    "fprep_data = fprep_closeup.get_data()[15:82, 72:106, :]\n",
    "nb.Nifti1Image(fprep_data, fprep_closeup.affine, fprep_closeup.header).to_filename('fprep_clip.nii.gz')\n",
    "\n",
    "\n",
    "msknii = nb.load(str(ATLAS_HOME / '1mm_tpm_csf.nii.gz'))\n",
    "fullbox = np.zeros(msknii.shape)\n",
    "fullbox[30:164, 142:213, 70:76] = 1\n",
    "fullbox[30:164, 142:213, 88:90] = 2\n",
    "fullbox[30:164, 142:213, 96:100] = 3\n",
    "nb.Nifti1Image(fullbox, msknii.affine, msknii.header).to_filename('box.nii.gz')\n",
    "\n",
    "newcsf = msknii.get_data()\n",
    "newcsf[:, :, 100:] = 0\n",
    "nb.Nifti1Image(newcsf, msknii.affine, msknii.header).to_filename('newcsfmsk.nii.gz')\n",
    "\n",
    "mskdat = msknii.get_data()[30:164, 142:213, :]\n",
    "nb.Nifti1Image(mskdat, msknii.affine, msknii.header).to_filename('csf_clip.nii.gz')\n",
    "msknii = nb.load(str(ATLAS_HOME / '1mm_tpm_wm.nii.gz'))\n",
    "mskdat = msknii.get_data()[30:164, 142:213, :]\n",
    "nb.Nifti1Image(mskdat, msknii.affine, msknii.header).to_filename('wm_clip.nii.gz')\n",
    "msknii = nb.load(str(ATLAS_HOME / '1mm_brainmask.nii.gz'))\n",
    "mskdat = msknii.get_data()[30:164, 142:213, :]\n",
    "nb.Nifti1Image(mskdat, msknii.affine, msknii.header).to_filename('bm_clip.nii.gz')\n",
    "\n",
    "mskbox = np.zeros_like(mskdat)\n",
    "mskbox[1:-1, 1:-1, :] = 1\n",
    "nb.Nifti1Image(mskbox, msknii.affine, msknii.header).to_filename('box_clip.nii.gz')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "colorboxes = [\n",
    "    (244/255, 3/255, 23/255),\n",
    "    (73/255, 213/255, 253/255),\n",
    "    (245/255, 29/255, 245/255)\n",
    "]\n",
    "fs = 25\n",
    "\n",
    "plt.clf()\n",
    "fig = plt.gcf()\n",
    "_ = fig.set_size_inches(15, 19.5)\n",
    "gs = gridspec.GridSpec(4, 2, width_ratios=[10, 1], height_ratios=[1, 1, 0.4, 3.7],  hspace=0., wspace=0.03)\n",
    "\n",
    "b_ax1 = plt.subplot(gs[0, 0])\n",
    "b_ax2 = plt.subplot(gs[1, 0])\n",
    "\n",
    "thres = 20\n",
    "vmin = 50\n",
    "vmax = 200\n",
    "\n",
    "disp = plotting.plot_anat(str(fprep_std), display_mode='z', annotate=False,\n",
    "                          cut_coords=[-15, -5, 10, 20, 40], cmap='cividis', threshold=thres, vmin=vmin, vmax=vmax,\n",
    "                          axes=b_ax1, resampling_interpolation='nearest')\n",
    "disp.annotate(size=15, left_right=True, positions=True)\n",
    "disp.add_contours('newcsfmsk.nii.gz', colors=['k'], levels=[0.8])\n",
    "disp.add_contours(str(ATLAS_HOME / '1mm_tpm_wm.nii.gz'), colors=['w'], levels=[0.8], linewidths=[1], alpha=0.7)\n",
    "disp.add_contours(str(ATLAS_HOME / '1mm_brainmask.nii.gz'), colors=['k'], levels=[0.8], linewidths=[3], alpha=.7)\n",
    "disp.add_contours('box.nii.gz', colors=colorboxes, levels=[0.8, 1.8, 2.8], linewidths=[3], alpha=1., zorder=20)\n",
    "\n",
    "\n",
    "disp = plotting.plot_anat('newfeat.nii.gz', display_mode='z', annotate=False,\n",
    "                          cut_coords=[-15, -5, 10, 20, 40], cmap='cividis', threshold=thres, vmin=vmin, vmax=vmax,\n",
    "                          axes=b_ax2, resampling_interpolation='nearest')\n",
    "disp.annotate(size=15, left_right=False, positions=False, scalebar=True, scale_width=2.5,\n",
    "              loc=3, size_vertical=2, label_top=False, frameon=True, borderpad=0.01)\n",
    "\n",
    "disp.add_contours('newcsfmsk.nii.gz', colors=['k'], levels=[0.8])\n",
    "disp.add_contours(str(ATLAS_HOME / '1mm_tpm_wm.nii.gz'), colors=['w'], levels=[0.8], linewidths=[1], alpha=0.7)\n",
    "disp.add_contours(str(ATLAS_HOME / '1mm_brainmask.nii.gz'), colors=['k'], levels=[0.8], linewidths=[3], alpha=.7)\n",
    "disp.add_contours('box.nii.gz', colors=colorboxes, levels=[0.8, 1.8, 2.8], linewidths=[3], alpha=1.)\n",
    "\n",
    "\n",
    "b_ax1.annotate(\n",
    "    'fMRIPrep',\n",
    "    xy=(0., .5), xycoords='axes fraction', xytext=(-fs, .0),\n",
    "    textcoords='offset points', va='center', color='k', size=fs,\n",
    "    rotation=90)\n",
    "\n",
    "b_ax2.annotate(\n",
    "    r'\\texttt{feat}',\n",
    "    xy=(0., .5), xycoords='axes fraction', xytext=(-fs, .0),\n",
    "    textcoords='offset points', va='center', color='k', size=fs,\n",
    "    rotation=90)\n",
    "\n",
    "\n",
    "inner_grid = gridspec.GridSpecFromSubplotSpec(1, 2, width_ratios=[1.5, 15],\n",
    "                                              subplot_spec=gs[:2, -1], wspace=0.01)\n",
    "\n",
    "b_ax3 = fig.add_subplot(inner_grid[0])\n",
    "gradient = np.hstack((np.zeros((50,)), np.linspace(0, 1, 120), np.ones((130,))))[::-1]\n",
    "gradient = np.vstack((gradient, gradient))\n",
    "b_ax3.imshow(gradient.T, aspect='auto', cmap=plt.get_cmap('cividis'))\n",
    "b_ax3.xaxis.set_ticklabels([])\n",
    "b_ax3.xaxis.set_ticks([])\n",
    "b_ax3.yaxis.set_ticklabels([])\n",
    "b_ax3.yaxis.set_ticks([])\n",
    "\n",
    "b_ax4 = fig.add_subplot(inner_grid[1])\n",
    "sn.distplot(nb.load(str(fprep_std)).get_data()[mask2mm], label='fMRIPrep',\n",
    "            vertical=True, ax=b_ax4, kde=False, norm_hist=True)\n",
    "sn.distplot(nb.load(str(feat_std)).get_data()[mask2mm], label=r'\\texttt{feat}', vertical=True,\n",
    "            color='darkorange', ax=b_ax4, kde=False, norm_hist=True)\n",
    "\n",
    "# plt.gca().set_ylim((0, 300))\n",
    "plt.legend(prop={'size': 15}, edgecolor='none')\n",
    "\n",
    "b_ax4.xaxis.set_ticklabels([])\n",
    "b_ax4.xaxis.set_ticks([])\n",
    "b_ax4.yaxis.set_ticklabels([])\n",
    "b_ax4.yaxis.set_ticks([])\n",
    "\n",
    "plt.axis('off')\n",
    "b_ax3.axis('off')\n",
    "b_ax4.axis('off')\n",
    "\n",
    "\n",
    "inner_grid = gridspec.GridSpecFromSubplotSpec(3, 2, width_ratios=[1, 1], hspace=0.04,\n",
    "                                              subplot_spec=gs[3, :], wspace=0.01)\n",
    "\n",
    "thres = 20\n",
    "vmin = 40\n",
    "vmax = 200\n",
    "\n",
    "\n",
    "for i, coord in enumerate([-5, 10, 20]):\n",
    "    ax1 = fig.add_subplot(inner_grid[i, 0])\n",
    "    disp = plotting.plot_anat(\n",
    "        'fprep_clip.nii.gz', display_mode='z', annotate=False,\n",
    "        cut_coords=[coord], cmap='cividis', threshold=thres, vmin=vmin, vmax=vmax,\n",
    "        resampling_interpolation='nearest', axes=ax1)\n",
    "    disp.add_contours('csf_clip.nii.gz', colors=['k'], levels=[0.8])\n",
    "    disp.add_contours('wm_clip.nii.gz', colors=['w'], levels=[0.8], linewidths=[1], alpha=0.7)\n",
    "    disp.add_contours('bm_clip.nii.gz', colors=['k'], levels=[0.8], linewidths=[3], alpha=.7)\n",
    "#     disp.annotate(size=42, left_right=False, positions=True, scalebar=False)\n",
    "    \n",
    "    ax1.axis('on')\n",
    "    ax1.set_xticklabels([])\n",
    "    ax1.set_xticks([])\n",
    "    ax1.set_yticklabels([])\n",
    "    ax1.set_yticks([])\n",
    "    ax1.set_ylabel(\"z = %d\" % coord, size=fs)\n",
    "    for pos in ['top', 'bottom', 'left', 'right']:\n",
    "        ax1.spines[pos].set_color(colorboxes[i])\n",
    "        ax1.spines[pos].set_visible(True)\n",
    "        ax1.spines[pos].set_linewidth(4)\n",
    "        ax1.spines[pos].set_position(('outward', 2))\n",
    "        \n",
    "    ax2 = fig.add_subplot(inner_grid[i, 1])\n",
    "    disp = plotting.plot_anat(\n",
    "        'feat_clip.nii.gz', display_mode='z', annotate=False,\n",
    "        cut_coords=[coord], cmap='cividis', threshold=thres, vmin=vmin, vmax=vmax,\n",
    "        resampling_interpolation='nearest', axes=ax2)\n",
    "    disp.add_contours('csf_clip.nii.gz', colors=['k'], levels=[0.8])\n",
    "    disp.add_contours('wm_clip.nii.gz', colors=['w'], levels=[0.8], linewidths=[1], alpha=0.7)\n",
    "    disp.add_contours('bm_clip.nii.gz', colors=['k'], levels=[0.8], linewidths=[3], alpha=.7)\n",
    "    \n",
    "    ax2.axis('on')\n",
    "    ax2.set_xticklabels([])\n",
    "    ax2.set_xticks([])\n",
    "    ax2.set_yticklabels([])\n",
    "    ax2.set_yticks([])\n",
    "    for pos in ['top', 'bottom', 'left', 'right']:\n",
    "        ax2.spines[pos].set_color(colorboxes[i])\n",
    "        ax2.spines[pos].set_visible(True)\n",
    "        ax2.spines[pos].set_linewidth(4)\n",
    "        ax2.spines[pos].set_position(('outward', 2))\n",
    "        \n",
    "    if i == 0:\n",
    "        ax1.set_title('fMRIPrep', size=fs+2, position=(0.5, 1.02))\n",
    "        ax2.set_title(r'\\texttt{feat}', size=fs+2, position=(0.5, 1.02))\n",
    "\n",
    "disp.annotate(size=32, left_right=False, positions=False, scalebar=True, scale_width=2.5,\n",
    "              loc=4, size_vertical=1.2, label_top=False, frameon=True, borderpad=0.2)\n",
    "\n",
    "ax1.annotate(\n",
    "    r'\\noindent\\parbox{25cm}{Variability mappings presented in this figure are derived from N=257 biologically independent participants}',\n",
    "    xy=(0.12, -0.15), xycoords='axes fraction', xytext=(.0, .0),\n",
    "    textcoords='offset points', va='center', color=(0.2, 0.2, 0.2), size=fs - 10,\n",
    ")\n",
    "\n",
    "plt.savefig(str(out_folder / 'SF04.pdf'),\n",
    "            format='pdf', bbox_inches='tight', pad_inches=0.2, dpi=300)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}