Deploy preview for PR 389 🛫

pr-preview/pr-389/.buildinfo

@@ -0,0 +1,4 @@
+# Sphinx build info version 1
+# This file hashes the configuration used when building these files. When it is not found, a full rebuild will be done.
+config: 8c982f8d80782740aefa675a008ebc5e
+tags: 645f666f9bcd5a90fca523b33c5a78b7

pr-preview/pr-389/_downloads/0129f8507efff39850cf153c51edb3ab/run_ets_rss_marker.py

@@ -0,0 +1,77 @@
+"""
+
+Extracting root sum of squares from edge-wise timeseries.
+=========================================================
+This example uses a ``RSSETSMarker`` to compute root sum of squares
+of the edge-wise timeseries using the Schaefer parcellation
+(100 rois and 200 rois, 17 Yeo networks) for a 4D nifti BOLD file.
+
+Authors: Leonard Sasse, Sami Hamdan, Nicolas Nieto, Synchon Mandal
+
+License: BSD 3 clause
+
+"""
+
+import tempfile
+
+import junifer.testing.registry  # noqa: F401
+from junifer.api import collect, run
+from junifer.storage import SQLiteFeatureStorage
+from junifer.utils import configure_logging
+
+
+###############################################################################
+# Set the logging level to info to see extra information:
+configure_logging(level="INFO")
+
+##############################################################################
+# Define the DataGrabber interface
+datagrabber = {
+    "kind": "SPMAuditoryTestingDataGrabber",
+}
+
+###############################################################################
+# Define the markers interface
+markers = [
+    {
+        "name": "Schaefer100x17_RSSETS",
+        "kind": "RSSETSMarker",
+        "parcellation": "Schaefer100x17",
+    },
+    {
+        "name": "Schaefer200x17_RSSETS",
+        "kind": "RSSETSMarker",
+        "parcellation": "Schaefer200x17",
+    },
+]
+
+###############################################################################
+# Create a temporary directory for junifer feature extraction:
+# At the end you can read the extracted data into a ``pandas.DataFrame``.
+with tempfile.TemporaryDirectory() as tmpdir:
+    # Define the storage interface
+    storage = {
+        "kind": "SQLiteFeatureStorage",
+        "uri": f"{tmpdir}/test.sqlite",
+    }
+    # Run the defined junifer feature extraction pipeline
+    run(
+        workdir=tmpdir,
+        datagrabber=datagrabber,
+        markers=markers,
+        storage=storage,
+        elements=["sub001"],  # we calculate for one subject only
+    )
+    # Collect extracted features data
+    collect(storage=storage)
+    # Create storage object to read in extracted features
+    db = SQLiteFeatureStorage(uri=storage["uri"])
+
+    # List all the features
+    print(db.list_features())
+    # Read extracted features
+    df_rssets = db.read_df(feature_name="BOLD_Schaefer200x17_RSSETS_rss_ets")
+
+###############################################################################
+# Now we take a look at the dataframe
+df_rssets.head()

pr-preview/pr-389/_downloads/042dd4fe6bffbb0ea320b529af464ace/run_run_gmd_mean.py

@@ -0,0 +1,53 @@
+"""
+UKB VBM GMD Extraction
+======================
+
+Authors: Federico Raimondo
+
+License: BSD 3 clause
+"""
+import tempfile
+
+import junifer.testing.registry  # noqa: F401
+from junifer.api import run
+
+
+datagrabber = {
+    "kind": "OasisVBMTestingDataGrabber",
+}
+
+markers = [
+    {
+        "name": "Schaefer1000x7_TrimMean80",
+        "kind": "ParcelAggregation",
+        "parcellation": "Schaefer1000x7",
+        "method": "trim_mean",
+        "method_params": {"proportiontocut": 0.2},
+    },
+    {
+        "name": "Schaefer1000x7_Mean",
+        "kind": "ParcelAggregation",
+        "parcellation": "Schaefer1000x7",
+        "method": "mean",
+    },
+    {
+        "name": "Schaefer1000x7_Std",
+        "kind": "ParcelAggregation",
+        "parcellation": "Schaefer1000x7",
+        "method": "std",
+    },
+]
+
+storage = {
+    "kind": "SQLiteFeatureStorage",
+}
+
+with tempfile.TemporaryDirectory() as tmpdir:
+    uri = f"{tmpdir}/test.sqlite"
+    storage["uri"] = uri
+    run(
+        workdir="/tmp",
+        datagrabber=datagrabber,
+        markers=markers,
+        storage=storage,
+    )

pr-preview/pr-389/_downloads/04a7e637e66cd7d7b36569a7cb7e70b5/norun_ukbvm_gmd.py

@@ -0,0 +1,43 @@
+"""
+UKB VBM GMD Extraction
+======================
+
+Authors: Federico Raimondo
+
+License: BSD 3 clause
+"""
+
+
+from junifer.api import run
+
+
+markers = [
+    {
+        "name": "Schaefer1000x7_TrimMean80",
+        "kind": "ParcelAggregation",
+        "parcellation": "Schaefer1000x7",
+        "method": "trim_mean",
+        "method_params": {"proportiontocut": 0.2},
+    },
+    {
+        "name": "Schaefer1000x7_Mean",
+        "kind": "ParcelAggregation",
+        "parcellation": "Schaefer1000x7",
+        "method": "mean",
+    },
+    {
+        "name": "Schaefer1000x7_Std",
+        "kind": "ParcelAggregation",
+        "parcellation": "Schaefer1000x7",
+        "method": "std",
+    },
+]
+
+run(
+    workdir="/tmp",
+    datagrabber="JuselessUKBVBM",
+    elements=("sub-1627474", "ses-2"),
+    markers=markers,
+    storage="SQLDataFrameStorage",
+    storage_params={"outpath": "/data/project/juniferexample"},
+)

pr-preview/pr-389/_downloads/35c7ed013e0aac743386dbe5f567501f/run_junifer_julearn.ipynb

@@ -0,0 +1,158 @@
+{
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "\n# Run junifer and julearn.\n\nThis example uses a ParcelAggregation marker to compute the mean of each parcel\nusing the Schaefer parcellation (100 rois, 7 Yeo networks) for a 3D nifti to\nextract some features for machine learning using julearn to predict some other\ndata.\n\nAuthors: Leonard Sasse, Sami Hamdan, Nicolas Nieto, Synchon Mandal\n\nLicense: BSD 3 clause\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "import tempfile\n\nimport nilearn\nimport pandas as pd\nfrom julearn import run_cross_validation, PipelineCreator\n\nimport junifer.testing.registry  # noqa: F401\nfrom junifer.api import collect, run\nfrom junifer.storage.sqlite import SQLiteFeatureStorage\nfrom junifer.utils import configure_logging"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "Set the logging level to info to see extra information:\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "configure_logging(level=\"INFO\")"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "Define the markers you want:\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "marker_dicts = [\n    {\n        \"name\": \"Schaefer100x17_TrimMean80\",\n        \"kind\": \"ParcelAggregation\",\n        \"parcellation\": \"Schaefer100x17\",\n        \"method\": \"trim_mean\",\n        \"method_params\": {\"proportiontocut\": 0.2},\n    },\n    {\n        \"name\": \"Schaefer200x17_Mean\",\n        \"kind\": \"ParcelAggregation\",\n        \"parcellation\": \"Schaefer200x17\",\n        \"method\": \"mean\",\n    },\n]"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "Define target and confounds for julearn machine learning:\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "y = \"age\"\nconfound = \"sex\""
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "Load the VBM phenotype data for machine learning data:\n- Fetch the Oasis dataset\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "oasis_dataset = nilearn.datasets.fetch_oasis_vbm()\nage = oasis_dataset.ext_vars[y][:10]\nsex = (\n    pd.Series(oasis_dataset.ext_vars[\"mf\"][:10])\n    .map(lambda x: 1 if x == \"F\" else 0)\n    .values\n)"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "Create a temporary directory for junifer feature extraction:\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "with tempfile.TemporaryDirectory() as tmpdir:\n    storage = {\"kind\": \"SQLiteFeatureStorage\", \"uri\": f\"{tmpdir}/test.sqlite\"}\n    # run the defined junifer feature extraction pipeline\n    run(\n        workdir=\"/tmp\",\n        datagrabber={\"kind\": \"OasisVBMTestingDataGrabber\"},\n        markers=marker_dicts,\n        storage=storage,\n    )\n\n    # read in extracted features and add confounds and targets\n    # for julearn run cross validation\n    collect(storage)\n    db = SQLiteFeatureStorage(uri=storage[\"uri\"])\n\n    df_vbm = db.read_df(feature_name=\"VBM_GM_Schaefer200x17_Mean_aggregation\")\n    oasis_subjects = [x[0] for x in df_vbm.index]\n    df_vbm.index = oasis_subjects"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "Using julearn for machine learning:\nWe predict the age given our vbm features and sex as a confound.\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "X = list(df_vbm.columns)\ndf_vbm[y] = age\ndf_vbm[confound] = sex\n\nX_types = {\n    \"features\": X,\n    \"confound\": confound,\n}\n\ncreator = PipelineCreator(problem_type=\"regression\", apply_to=\"features\")\ncreator.add(\"zscore\", apply_to=[\"features\", \"confound\"])\ncreator.add(\"confound_removal\", apply_to=\"features\", confounds=\"confound\")\ncreator.add(\"ridge\")\n\nscores = run_cross_validation(\n    X=X + [confound],\n    y=y,\n    X_types=X_types,\n    data=df_vbm,\n    model=creator,\n    cv=3,\n)\nprint(scores)"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "Interpretation of results:\nDoing machine learning with only 10 datapoints is not meaningful.\nThis explains the big variation in scores\nfor different cross-validation folds.\n\n"
+      ]
+    }
+  ],
+  "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": "3.12.7"
+    }
+  },
+  "nbformat": 4,
+  "nbformat_minor": 0
+}