"""

.. _datasets_neurovault:

=========================================
Use NeuroVault statistical maps in NiMARE
=========================================

Download statistical maps from NeuroVault, then use them in a meta-analysis,
with NiMARE.
"""
import matplotlib.pyplot as plt
from nilearn.plotting import plot_stat_map

###############################################################################
# Neurovault + NiMARE: Load freely shared statistical maps for Meta-Analysis
# -----------------------------------------------------------------------------
# `Neurovault <https://neurovault.org/>`_ is an online platform that hosts
# unthresholded statistical maps, including group statistical maps.
# NiMARE can read these statistical maps when given a list of collection_ids.
# I search "working memory" on neurovault, and find these relevant collections:
#
# * `2884 <https://neurovault.org/collections/2884/>`_
# * `2621 <https://neurovault.org/collections/2621/>`_
# * `3085 <https://neurovault.org/collections/3085/>`_
# * `5623 <https://neurovault.org/collections/5623/>`_
# * `3264 <https://neurovault.org/collections/3264/>`_
# * `3192 <https://neurovault.org/collections/3192/>`_
# * `457 <https://neurovault.org/collections/457/>`_
#
# I can load specific statistical maps from these collections
# directly into a Studyset for analysis:
from nimare.generate import create_neurovault_studyset

# The specific collections I would like to download group level
# statistical maps from
collection_ids = (2884, 2621, 3085, 5623, 3264, 3192, 457)

# A mapping between what I want the contrast(s) to be
# named in the Studyset and what their respective group
# statistical maps are named on neurovault
contrasts = {
    "working_memory": (
        "Working memory load of 2 faces versus 1 face - NT2_Tstat|"
        "t-value contrast 2-back minus 0-back|"
        "Searchlight multivariate Decoding 2: visual working memory|"
        "Context-dependent group-specific WM information|"
        "WM working memory zstat1|"
        "WM task over CRT task map|"
        "tfMRI WM 2BK PLACE zstat1"
    )
}

# Convert how the statistical maps on neurovault are represented
# in a NiMARE Studyset.
map_type_conversion = {"Z map": "z", "T map": "t"}

studyset = create_neurovault_studyset(
    collection_ids,
    contrasts,
    img_dir=None,
    map_type_conversion=map_type_conversion,
)

###############################################################################
# Conversion of Statistical Maps
# -----------------------------------------------------------------------------
# ``create_neurovault_studyset`` already resolves compatible image types.
# To explicitly demonstrate :class:`~nimare.transforms.ImageTransformer` on a
# Studyset-backed collection, we drop the derived Z maps from contrasts that
# still have T maps and regenerate them.
from nimare.transforms import ImageTransformer

images = studyset.images.copy()
images.loc[images["t"].notnull(), "z"] = None
studyset.images = images

# Some studies are now missing Z maps again.
studyset.images[["t", "z"]]

###############################################################################
z_transformer = ImageTransformer(target="z")
studyset = z_transformer.transform(studyset)

###############################################################################
# All studies now have Z maps again.
studyset.images[["z"]]

###############################################################################
# Run a Meta-Analysis
# -----------------------------------------------------------------------------
# With the missing Z maps filled in, we can run a Meta-Analysis
# and plot our results
from nimare.meta.ibma import Fishers

# The default template has a slightly different, but completely compatible,
# affine than the NeuroVault images, so we allow the Estimator to resample
# images during the fitting process.
meta = Fishers(resample=True)

# Drop studies that have no Z map (e.g. collections with no downloadable images).
has_z = studyset.images["z"].notnull()
valid_ids = studyset.images.loc[has_z, "id"].tolist()
if valid_ids:
    studyset = studyset.filter_ids(valid_ids)

meta_res = meta.fit(studyset)

fig, ax = plt.subplots()
display = plot_stat_map(meta_res.get_map("z"), threshold=3.3, axes=ax, figure=fig)
fig.show()
# The result may look questionable, but this code provides
# a template on how to use neurovault in your meta analysis.