Merge branch 'main' of github.com:SpikeInterface/spikeinterface into …

…rel-path-phy

doc/how_to/index.rst

@@ -7,3 +7,4 @@ How to guides
     get_started
     analyse_neuropixels
     handle_drift
+    load_matlab_data

doc/how_to/load_matlab_data.rst

@@ -0,0 +1,100 @@
+Exporting MATLAB Data to Binary & Loading in SpikeInterface
+===========================================================
+
+In this tutorial, we will walk through the process of exporting data from MATLAB in a binary format and subsequently loading it using SpikeInterface in Python.
+
+Exporting Data from MATLAB
+--------------------------
+
+Begin by ensuring your data structure is correct. Organize your data matrix so that the first dimension corresponds to samples/time and the second to channels.
+Here, we present a MATLAB code that creates a random dataset and writes it to a binary file as an illustration.
+
+.. code-block:: matlab
+
+   % Define the size of your data
+   numSamples = 1000;
+   numChannels = 384;
+
+   % Generate random data as an example
+   data = rand(numSamples, numChannels);
+
+   % Write the data to a binary file
+   fileID = fopen('your_data_as_a_binary.bin', 'wb');
+   fwrite(fileID, data, 'double');
+   fclose(fileID);
+
+.. note::
+
+   In your own script, replace the random data generation with your actual dataset.
+
+Loading Data in SpikeInterface
+------------------------------
+
+After executing the above MATLAB code, a binary file named `your_data_as_a_binary.bin` will be created in your MATLAB directory. To load this file in Python, you'll need its full path.
+
+Use the following Python script to load the binary data into SpikeInterface:
+
+.. code-block:: python
+
+   import spikeinterface as si
+   from pathlib import Path
+
+   # Define file path
+   # For Linux or macOS:
+   file_path = Path("/The/Path/To/Your/Data/your_data_as_a_binary.bin")
+   # For Windows:
+   # file_path = Path(r"c:\path\to\your\data\your_data_as_a_binary.bin")
+
+   # Confirm file existence
+   assert file_path.is_file(), f"Error: {file_path} is not a valid file. Please check the path."
+
+   # Define recording parameters
+   sampling_frequency = 30_000.0  # Adjust according to your MATLAB dataset
+   num_channels = 384  # Adjust according to your MATLAB dataset
+   dtype = "float64"  # MATLAB's double corresponds to Python's float64
+
+   # Load data using SpikeInterface
+   recording = si.read_binary(file_path, sampling_frequency=sampling_frequency,
+                                        num_channels=num_channels, dtype=dtype)
+
+   # Confirm that the data was loaded correctly by comparing the data shapes and see they match the MATLAB data
+   print(recording.get_num_frames(), recording.get_num_channels())
+
+Follow the steps above to seamlessly import your MATLAB data into SpikeInterface. Once loaded, you can harness the full power of SpikeInterface for data processing, including filtering, spike sorting, and more.
+
+Common Pitfalls & Tips
+----------------------
+
+1. **Data Shape**: Make sure your MATLAB data matrix's first dimension is samples/time and the second is channels. If your time is in the second dimension, use `time_axis=1` in `si.read_binary()`.
+2. **File Path**: Always double-check the Python file path.
+3. **Data Type Consistency**: Ensure data types between MATLAB and Python are consistent. MATLAB's `double` is equivalent to Numpy's `float64`.
+4. **Sampling Frequency**: Set the appropriate sampling frequency in Hz for SpikeInterface.
+5. **Transition to Python**: Moving from MATLAB to Python can be challenging. For newcomers to Python, consider reviewing numpy's [Numpy for MATLAB Users](https://numpy.org/doc/stable/user/numpy-for-matlab-users.html) guide.
+
+Using gains and offsets for integer data
+----------------------------------------
+
+Raw data formats often store data as integer values for memory efficiency. To give these integers meaningful physical units, you can apply a gain and an offset.
+In SpikeInterface, you can use the `gain_to_uV` and `offset_to_uV` parameters, since traces are handled in microvolts (uV). Both parameters can be integrated into the `read_binary` function.
+If your data in MATLAB is stored as `int16`, and you know the gain and offset, you can use the following code to load the data:
+
+.. code-block:: python
+
+   sampling_frequency = 30_000.0  # Adjust according to your MATLAB dataset
+   num_channels = 384  # Adjust according to your MATLAB dataset
+   dtype_int = 'int16'  # Adjust according to your MATLAB dataset
+   gain_to_uV = 0.195  # Adjust according to your MATLAB dataset
+   offset_to_uV = 0   # Adjust according to your MATLAB dataset
+
+   recording = si.read_binary(file_path, sampling_frequency=sampling_frequency,
+                              num_channels=num_channels, dtype=dtype_int,
+                              gain_to_uV=gain_to_uV, offset_to_uV=offset_to_uV)
+
+   recording.get_traces(return_scaled=True)  # Return traces in micro volts (uV)
+
+
+This will equip your recording object with capabilities to convert the data to float values in uV using the :code:`get_traces()` method with the :code:`return_scaled` parameter set to :code:`True`.
+
+.. note::
+
+   The gain and offset parameters are usually format dependent and you will need to find out the correct values for your data format. You can load your data without gain and offset but then the traces will be in integer values and not in uV.

doc/modules/core.rst

@@ -547,8 +547,7 @@ workflow.
 In order to do this, one can use the :code:`Numpy*` classes, :py:class:`~spikeinterface.core.NumpyRecording`,
 :py:class:`~spikeinterface.core.NumpySorting`, :py:class:`~spikeinterface.core.NumpyEvent`, and
 :py:class:`~spikeinterface.core.NumpySnippets`. These object behave exactly like normal SpikeInterface objects,
-but they are not bound to a file. This makes these objects *not dumpable*, so parallel processing is not supported.
-In order to make them *dumpable*, one can simply :code:`save()` them (see :ref:`save_load`).
+but they are not bound to a file.
 
 Also note the class :py:class:`~spikeinterface.core.SharedMemorySorting` which is very similar to
 Similar to :py:class:`~spikeinterface.core.NumpySorting` but with an unerlying SharedMemory which is usefull for

doc/modules/qualitymetrics.rst

@@ -25,9 +25,11 @@ For more details about each metric and it's availability and use within SpikeInt
   :glob:
 
   qualitymetrics/amplitude_cutoff
+  qualitymetrics/amplitude_cv
   qualitymetrics/amplitude_median
   qualitymetrics/d_prime
   qualitymetrics/drift
+  qualitymetrics/firing_range
   qualitymetrics/firing_rate
   qualitymetrics/isi_violations
   qualitymetrics/isolation_distance

doc/modules/qualitymetrics/amplitude_cutoff.rst

@@ -21,12 +21,12 @@ Example code
 
 .. code-block:: python
 
-	import spikeinterface.qualitymetrics as qm
+	import spikeinterface.qualitymetrics as sqm
 
 	# It is also recommended to run `compute_spike_amplitudes(wvf_extractor)`
 	# in order to use amplitudes from all spikes
-	fraction_missing = qm.compute_amplitude_cutoffs(wvf_extractor, peak_sign="neg")
-	# fraction_missing is a dict containing the units' IDs as keys,
+	fraction_missing = sqm.compute_amplitude_cutoffs(wvf_extractor, peak_sign="neg")
+	# fraction_missing is a dict containing the unit IDs as keys,
 	# and their estimated fraction of missing spikes as values.
 
 Reference

doc/modules/qualitymetrics/amplitude_cv.rst

@@ -0,0 +1,55 @@
+Amplitude CV (:code:`amplitude_cv_median`, :code:`amplitude_cv_range`)
+======================================================================
+
+
+Calculation
+-----------
+
+The amplitude CV (coefficient of variation) is a measure of the amplitude variability.
+It is computed as the ratio between the standard deviation and the amplitude mean.
+To obtain a better estimate of this measure, it is first computed separately for several temporal bins.
+Out of these values, the median and the range (percentile distance, by default between the
+5th and 95th percentiles) are computed.
+
+The computation requires either spike amplitudes (see :py:func:`~spikeinterface.postprocessing.compute_spike_amplitudes()`)
+or amplitude scalings (see :py:func:`~spikeinterface.postprocessing.compute_amplitude_scalings()`) to be pre-computed.
+
+
+Expectation and use
+-------------------
+
+The amplitude CV median is expected to be relatively low for well-isolated units, indicating a "stereotypical" spike shape.
+
+The amplitude CV range can be high in the presence of noise contamination, due to amplitude outliers like in
+the example below.
+
+.. image:: amplitudes.png
+    :width: 600
+
+
+Example code
+------------
+
+.. code-block:: python
+
+    import spikeinterface.qualitymetrics as sqm
+
+    # Make recording, sorting and wvf_extractor object for your data.
+	# It is required to run `compute_spike_amplitudes(wvf_extractor)` or
+	# `compute_amplitude_scalings(wvf_extractor)` (if missing, values will be NaN)
+    amplitude_cv_median, amplitude_cv_range = sqm.compute_amplitude_cv_metrics(wvf_extractor)
+    # amplitude_cv_median and  amplitude_cv_range are dicts containing the unit ids as keys,
+    # and their amplitude_cv metrics as values.
+
+
+
+References
+----------
+
+.. autofunction:: spikeinterface.qualitymetrics.misc_metrics.compute_amplitude_cv_metrics
+
+
+Literature
+----------
+
+Designed by Simon Musall and adapted to SpikeInterface by Alessio Buccino.

doc/modules/qualitymetrics/amplitude_median.rst

@@ -20,12 +20,12 @@ Example code
 
 .. code-block:: python
 
-	import spikeinterface.qualitymetrics as qm
+	import spikeinterface.qualitymetrics as sqm
 
 	# It is also recommended to run `compute_spike_amplitudes(wvf_extractor)`
 	# in order to use amplitude values from all spikes.
-	amplitude_medians = qm.compute_amplitude_medians(wvf_extractor)
-	# amplitude_medians is a dict containing the units' IDs as keys,
+	amplitude_medians = sqm.compute_amplitude_medians(wvf_extractor)
+	# amplitude_medians is a dict containing the unit IDs as keys,
 	# and their estimated amplitude medians as values.
 
 Reference

doc/modules/qualitymetrics/d_prime.rst

@@ -32,9 +32,9 @@ Example code
 
 .. code-block:: python
 
-    import spikeinterface.qualitymetrics as qm
+    import spikeinterface.qualitymetrics as sqm
 
-    d_prime = qm.lda_metrics(all_pcs, all_labels, 0)
+    d_prime = sqm.lda_metrics(all_pcs, all_labels, 0)
 
 
 Reference

doc/modules/qualitymetrics/drift.rst

@@ -40,11 +40,12 @@ Example code
 
 .. code-block:: python
 
-	import spikeinterface.qualitymetrics as qm
+	import spikeinterface.qualitymetrics as sqm
 
+	# Make recording, sorting and wvf_extractor object for your data.
 	# It is required to run `compute_spike_locations(wvf_extractor)`
 	# (if missing, values will be NaN)
-	drift_ptps, drift_stds, drift_mads = qm.compute_drift_metrics(wvf_extractor, peak_sign="neg")
+	drift_ptps, drift_stds, drift_mads = sqm.compute_drift_metrics(wvf_extractor, peak_sign="neg")
 	# drift_ptps, drift_stds, and drift_mads are dict containing the units' ID as keys,
 	# and their metrics as values.
 

doc/modules/qualitymetrics/firing_range.rst

@@ -0,0 +1,40 @@
+Firing range (:code:`firing_range`)
+===================================
+
+
+Calculation
+-----------
+
+The firing range indicates the dispersion of the firing rate of a unit across the recording. It is computed by
+taking the difference between the 95th percentile's firing rate and the 5th percentile's firing rate computed over short time bins (e.g. 10 s).
+
+
+
+Expectation and use
+-------------------
+
+Very high levels of firing ranges, outside of a physiological range, might indicate noise contamination.
+
+
+Example code
+------------
+
+.. code-block:: python
+
+    import spikeinterface.qualitymetrics as sqm
+
+    # Make recording, sorting and wvf_extractor object for your data.
+    firing_range = sqm.compute_firing_ranges(wvf_extractor)
+    # firing_range is a dict containing the unit IDs as keys,
+    # and their firing firing_range as values (in Hz).
+
+References
+----------
+
+.. autofunction:: spikeinterface.qualitymetrics.misc_metrics.compute_firing_ranges
+
+
+Literature
+----------
+
+Designed by Simon Musall and adapted to SpikeInterface by Alessio Buccino.

doc/modules/qualitymetrics/firing_rate.rst

@@ -37,11 +37,11 @@ With SpikeInterface:
 
 .. code-block:: python
 
-    import spikeinterface.qualitymetrics as qm
+    import spikeinterface.qualitymetrics as sqm
 
     # Make recording, sorting and wvf_extractor object for your data.
-    firing_rate = qm.compute_firing_rates(wvf_extractor)
-    # firing_rate is a dict containing the units' IDs as keys,
+    firing_rate = sqm.compute_firing_rates(wvf_extractor)
+    # firing_rate is a dict containing the unit IDs as keys,
     # and their firing rates across segments as values (in Hz).
 
 References

doc/modules/qualitymetrics/isi_violations.rst

@@ -77,11 +77,11 @@ With SpikeInterface:
 
 .. code-block:: python
 
-    import spikeinterface.qualitymetrics as qm
+    import spikeinterface.qualitymetrics as sqm
 
     # Make recording, sorting and wvf_extractor object for your data.
 
-    isi_violations_ratio, isi_violations_count = qm.compute_isi_violations(wvf_extractor, isi_threshold_ms=1.0)
+    isi_violations_ratio, isi_violations_count = sqm.compute_isi_violations(wvf_extractor, isi_threshold_ms=1.0)
 
 References
 ----------

doc/modules/qualitymetrics/presence_ratio.rst

@@ -23,12 +23,12 @@ Example code
 
 .. code-block:: python
 
-    import spikeinterface.qualitymetrics as qm
+    import spikeinterface.qualitymetrics as sqm
 
     # Make recording, sorting and wvf_extractor object for your data.
 
-    presence_ratio = qm.compute_presence_ratios(wvf_extractor)
-    # presence_ratio is a dict containing the units' IDs as keys
+    presence_ratio = sqm.compute_presence_ratios(wvf_extractor)
+    # presence_ratio is a dict containing the unit IDs as keys
     # and their presence ratio (between 0 and 1) as values.
 
 Links to original implementations

doc/modules/qualitymetrics/sliding_rp_violations.rst

@@ -27,11 +27,11 @@ With SpikeInterface:
 
 .. code-block:: python
 
-    import spikeinterface.qualitymetrics as qm
+    import spikeinterface.qualitymetrics as sqm
 
     # Make recording, sorting and wvf_extractor object for your data.
 
-    contamination = qm.compute_sliding_rp_violations(wvf_extractor, bin_size_ms=0.25)
+    contamination = sqm.compute_sliding_rp_violations(wvf_extractor, bin_size_ms=0.25)
 
 References
 ----------

doc/modules/qualitymetrics/snr.rst

@@ -41,12 +41,12 @@ With SpikeInterface:
 
 .. code-block:: python
 
-    import spikeinterface.qualitymetrics as qm
+    import spikeinterface.qualitymetrics as sqm
 
     # Make recording, sorting and wvf_extractor object for your data.
 
-    SNRs = qm.compute_snrs(wvf_extractor)
-    # SNRs is a dict containing the units' IDs as keys and their SNRs as values.
+    SNRs = sqm.compute_snrs(wvf_extractor)
+    # SNRs is a dict containing the unit IDs as keys and their SNRs as values.
 
 Links to original implementations
 ---------------------------------

doc/modules/qualitymetrics/synchrony.rst

@@ -27,9 +27,9 @@ Example code
 
 .. code-block:: python
 
-    import spikeinterface.qualitymetrics as qm
+    import spikeinterface.qualitymetrics as sqm
     # Make recording, sorting and wvf_extractor object for your data.
-    synchrony = qm.compute_synchrony_metrics(wvf_extractor, synchrony_sizes=(2, 4, 8))
+    synchrony = sqm.compute_synchrony_metrics(wvf_extractor, synchrony_sizes=(2, 4, 8))
     # synchrony is a tuple of dicts with the synchrony metrics for each unit