funcnodes-span 0.1.2__tar.gz → 0.1.4__tar.gz

{funcnodes_span-0.1.2 → funcnodes_span-0.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: funcnodes-span
-Version: 0.1.2
+Version: 0.1.4
 Summary: 
 Author: Kourosh Rezaei
 Author-email: kouroshrezaei90@gmail.com
@@ -11,6 +11,8 @@ Classifier: Programming Language :: Python :: 3.12
 Requires-Dist: funcnodes (>=0.2.21,<0.3.0)
 Requires-Dist: funcnodes_numpy (>=0.1.57,<0.2.0)
 Requires-Dist: funcnodes_pandas (>=0.1.9,<0.2.0)
+Requires-Dist: funcnodes_plotly (>=0.1.7,<0.2.0)
+Requires-Dist: lmfit (>=1.3.2,<2.0.0)
 Requires-Dist: pooch (>=1.8.2,<2.0.0)
 Requires-Dist: scipy (>=1.14.0,<2.0.0)
 Description-Content-Type: text/markdown

{funcnodes_span-0.1.2 → funcnodes_span-0.1.4}/funcnodes_span/__init__.py

@@ -2,9 +2,9 @@ import funcnodes as fn
 
 from .normalization import NORM_NODE_SHELF as NORM
 from .smoothing import SMOOTH_NODE_SHELF as SMOOTH
+from .peak_analysis import PEAK_NODE_SHELF as PEAK
 
-
-__version__ = "0.1.2"
+__version__ = "0.1.4"
 
 NODE_SHELF = fn.Shelf(
     name="Spectral Analysis",
@@ -13,5 +13,6 @@ NODE_SHELF = fn.Shelf(
     subshelves=[
         NORM,
         SMOOTH,
+        PEAK
     ],
 )

funcnodes_span-0.1.4/funcnodes_span/peak_analysis.py

@@ -0,0 +1,556 @@
+from funcnodes import NodeDecorator, Shelf
+import numpy as np
+from enum import Enum
+from exposedfunctionality import controlled_wrapper
+from typing import Optional, TypedDict, List
+from scipy.signal import find_peaks
+from scipy.stats import norm
+from scipy import interpolate
+import copy
+import lmfit
+import plotly.graph_objs as go
+from plotly.subplots import make_subplots
+import re
+
+
+class PeakProperties(TypedDict):
+    id: str
+    i_index: int
+    index: int
+    f_index: int
+    x_at_i_index: int
+    x_at_index: int
+    x_at_f_index: int
+    y_at_index: int
+    y_at_f_index: int
+    y_at_i_index: int
+    area: float
+    symmetricity: float
+    tailing: float
+    FWHM: float
+    plate_nr: float
+    width: float
+    _is_fitted: bool = False
+    _is_force_fitted: bool = False
+    fitting_data: Optional[dict] = None
+    fitting_info: Optional[dict] = None
+
+
+def compute_peak_properties(
+    x_array: np.ndarray,
+    y_array: np.ndarray,
+    peak_indices: List[int],
+    peak_nr: int,
+    is_fitted: bool = False,
+    is_force_fitted: bool = False,
+    fitting_data: Optional[dict] = None,
+    fitting_info: Optional[dict] = None,
+) -> PeakProperties:
+    # """
+    # Compute various properties of a given peak.
+
+    # Parameters:
+    # - x_array: np.ndarray - The array of x-values (e.g., time or wavelength).
+    # - y_array: np.ndarray - The array of y-values (e.g., intensity).
+    # - peak_indices: List[int] - A list containing the start index, peak index, and end index of the peak.
+    # - peak_nr: int - The identifier number of the peak.
+    # - is_fitted: bool = False - A flag indicating whether the peak is fitted or not.
+    # - is_force_fitted: bool = False - A flag indicating whether the peak is forced fitted or not.
+    # - fitting_data: Optional[dict] = None - A dictionary containing the fitting data if the peak is fitted.
+    # - fitting_info: Optional[dict] = None - A dictionary containing the fitting information if the peak is fitted.
+
+    # Returns:
+    # - peak_properties: PeakProperties - A dictionary containing various properties of the peak.
+    # """
+
+    i_index, index, f_index = peak_indices
+
+    # Extract the relevant portion of the arrays
+    selected_signal = y_array[i_index:f_index]
+    selected_time = x_array[i_index:f_index]
+
+    # Create an interpolated time array with higher resolution
+    selected_time_interpol = np.linspace(
+        selected_time[0],
+        selected_time[-1],
+        num=len(selected_time) * 10,
+        endpoint=True,
+    )
+
+    # Interpolate the selected signal to match the interpolated time array
+    f_interpol = interpolate.interp1d(selected_time, selected_signal, kind="linear")
+    selected_signal_interpol = f_interpol(selected_time_interpol)
+
+    # Determine the amplitude and position of the peak in the interpolated signal
+    amplitude = np.max(selected_signal_interpol)
+    peak_position = np.where(selected_signal_interpol == amplitude)[0][0]
+
+    # Split the interpolated signal into left and right spectra relative to the peak
+    left_spectrum = selected_signal_interpol[:peak_position]
+    right_spectrum = selected_signal_interpol[peak_position:]
+
+    # Helper function to compute indices for a given percentage of amplitude
+    def compute_indices(spectrum, percentage):
+        try:
+            left_index = (np.abs(spectrum - percentage * amplitude)).argmin()
+        except ValueError:
+            left_index = 0
+        return left_index
+
+    # Compute FWHM (Full Width at Half Maximum)
+    FWHM_left_index = compute_indices(left_spectrum, 0.5)
+    FWHM_right_index = compute_indices(right_spectrum, 0.5) + peak_position
+    FWHM_a = abs(
+        selected_time_interpol[FWHM_left_index] - selected_time_interpol[peak_position]
+    )
+    FWHM_b = abs(
+        selected_time_interpol[FWHM_right_index] - selected_time_interpol[peak_position]
+    )
+    FWHM = np.around(FWHM_b / FWHM_a, 2) if FWHM_a != 0 else np.nan
+
+    # Compute Symmetricity
+    symmetricity_left_index = compute_indices(left_spectrum, 0.1)
+    symmetricity_right_index = compute_indices(right_spectrum, 0.1) + peak_position
+    symmetricity_a = abs(
+        selected_time_interpol[symmetricity_left_index]
+        - selected_time_interpol[peak_position]
+    )
+    symmetricity_b = abs(
+        selected_time_interpol[symmetricity_right_index]
+        - selected_time_interpol[peak_position]
+    )
+    symmetricity = (
+        np.around(symmetricity_b / symmetricity_a, 2) if symmetricity_a != 0 else np.nan
+    )
+
+    # Compute Tailing
+    tailing_left_index = compute_indices(left_spectrum, 0.05)
+    tailing_right_index = compute_indices(right_spectrum, 0.05) + peak_position
+    tailing_a = abs(
+        selected_time_interpol[tailing_left_index]
+        - selected_time_interpol[peak_position]
+    )
+    tailing_b = abs(
+        selected_time_interpol[tailing_right_index]
+        - selected_time_interpol[peak_position]
+    )
+    tailing = (
+        np.around(((tailing_a + tailing_b) / 2 * tailing_a), 2)
+        if tailing_a != 0
+        else np.nan
+    )
+
+    # Compute Area under the peak
+    area = abs(np.trapz(selected_signal_interpol, selected_time_interpol))
+
+    # Compute Plate Number
+    try:
+        plate_nr = 2 * np.pi * ((x_array[i_index] * y_array[index]) / area) ** 2
+    except ZeroDivisionError:
+        plate_nr = np.nan
+
+    # Compute Width of the peak
+    width = x_array[f_index] - x_array[i_index]
+
+    # Populate the PeakProperties dictionary
+    peak_properties: PeakProperties = {
+        "id": str(peak_nr + 1) + "_fitted" if is_fitted else str(peak_nr + 1),
+        "i_index": i_index,
+        "index": index,
+        "f_index": f_index,
+        "x_at_i_index": x_array[i_index],
+        "x_at_index": x_array[index],
+        "x_at_f_index": x_array[f_index],
+        "y_at_i_index": y_array[i_index],
+        "y_at_index": y_array[index],
+        "y_at_f_index": y_array[f_index],
+        "area": area,
+        "symmetricity": symmetricity,
+        "tailing": tailing,
+        "FWHM": FWHM,
+        "plate_nr": plate_nr,
+        "width": width,
+        "_is_fitted": is_fitted,
+        "_is_force_fitted": is_force_fitted,
+        "fitting_data": fitting_data,
+        "fitting_info": fitting_info,
+    }
+
+    return peak_properties
+
+
+@NodeDecorator(id="span.basics.peaks", name="Peak finder node")
+@controlled_wrapper(find_peaks, wrapper_attribute="__fnwrapped__")
+def peak_finder(
+    x_array: np.ndarray,
+    y_array: np.ndarray,
+    noise_level: Optional[int] = None,
+    height: Optional[float] = None,
+    threshold: Optional[float] = None,
+    distance: Optional[float] = None,
+    prominence: Optional[float] = None,
+    width: Optional[float] = None,
+    wlen: Optional[int] = None,
+    rel_height: Optional[float] = None,
+    plateau_size: Optional[int] = None,
+) -> dict:
+    peak_lst = []
+    height = 0.05 * np.max(y_array) if height is None else height
+    noise_level = 5000 if noise_level is None else noise_level
+
+    # Make a copy of the input array
+    y_array_copy = np.copy(y_array)
+
+    # Find the peaks in the copy of the input array
+    peaks, _ = find_peaks(
+        y_array_copy,
+        threshold=threshold,
+        prominence=prominence,
+        height=height,
+        distance=distance,
+        width=width,
+        wlen=wlen,
+        rel_height=rel_height,
+        plateau_size=plateau_size,
+    )
+
+    # Calculate the standard deviation of peak prominences
+
+    np.random.seed(seed=1)
+    # Fit a normal distribution to the input array
+    mu, std = norm.fit(y_array_copy)
+    if peaks is not None:
+        try:
+            # Add noise to the input array
+            noise = np.random.normal(
+                mu / noise_level, std / noise_level, np.shape(y_array_copy)
+            )
+            y_array_copy = y_array_copy + noise
+
+            # Find the minimums in the copy of the input array
+            mins, _ = find_peaks(-1 * y_array_copy)
+
+            # Iterate over the peaks
+            for peak in peaks:
+                # Calculate the prominences of the peak
+                # Find the right minimum of the peak
+                right_min = mins[np.argmax(mins > peak)]
+                if right_min < peak:
+                    right_min = len(y_array) - 1
+
+                try:
+                    # Find the left minimum of the peak
+                    left_min = np.array(mins)[np.where(np.array(mins) < peak)][-1]
+                except IndexError:
+                    left_min = 0
+
+                if height is None:
+                    # If no height is specified, append the peak bounds to the peak list
+                    peak_lst.append([left_min, peak, right_min])
+
+                else:
+                    # If a height is specified, append the peak bounds to the peak list
+                    # if the peak's value is greater than the height
+                    if y_array_copy[peak] > height:
+                        peak_lst.append([left_min, peak, right_min])
+
+        except ValueError:
+            # If an error occurs when adding noise to the input array, add stronger noise and try again
+            noise = np.random.normal(mu / 100, std / 100, np.shape(y_array_copy))
+            y_array_copy = y_array_copy + noise
+            mins, _ = find_peaks(-1 * y_array_copy)
+            for peak in peaks:
+                right_min = mins[np.argmax(mins > peak)]
+                if right_min < peak:
+                    right_min = len(y_array) - 1
+                try:
+                    left_min = np.array(mins)[np.where(np.array(mins) < peak)][-1]
+                except IndexError:
+                    left_min = 0
+                if height is None:
+                    # If no height is specified, append the peak bounds to the peak list
+                    peak_lst.append([left_min, peak, right_min])
+                else:
+                    # If a height is specified, append the peak bounds to the peak list
+                    # if the peak's value is greater than the height
+                    if y_array_copy[peak] > height:
+                        peak_lst.append([left_min, peak, right_min])
+
+    peak_properties_list = []
+
+    for peak_nr, peak in enumerate(peak_lst):
+        peak_properties = compute_peak_properties(
+            x_array=x_array, y_array=y_array, peak_indices=peak, peak_nr=peak_nr
+        )
+        peak_properties_list.append(peak_properties)
+
+    return peak_properties_list
+
+
+# ['Constant', 'Complex Constant', 'Linear', 'Quadratic', 'Polynomial',
+# 'Spline', 'Gaussian', 'Gaussian-2D', 'Lorentzian', 'Split-Lorentzian', 'Voigt',
+# 'PseudoVoigt', 'Moffat', 'Pearson4', 'Pearson7', 'StudentsT', 'Breit-Wigner', 'Log-Normal',
+# 'Damped Oscillator', 'Damped Harmonic Oscillator', 'Exponential Gaussian', 'Skewed Gaussian',
+# 'Skewed Voigt', 'Thermal Distribution', 'Doniach', 'Power Law', 'Exponential', 'Step',
+# 'Rectangle', 'Expression']
+
+
+class FittingModel(Enum):
+    ComplexConstant = "Complex Constant"
+    Linear = "Linear"
+    Quadratic = "Quadratic"
+    Polynomial = "Polynomial"
+    Spline = "Spline"
+    Gaussian = "Gaussian"
+    Gaussian2D = "Gaussian-2D"
+    Lorentzian = "Lorentzian"
+    SplitLorentzian = "Split-Lorentzian"
+    Voigt = "Voigt"
+    PseudoVoigt = "PseudoVoigt"
+    Moffat = "Moffat"
+    Pearson4 = "Pearson4"
+    Pearson7 = "Pearson7"
+    StudentsT = "StudentsT"
+    BreitWigner = "Breit-Wigner"
+    LogNormal = "Log-Normal"
+    DampedOscillator = "Damped Oscillator"
+    DampedHarmonicOscillator = "Damped Harmonic Oscillator"
+    ExponentialGaussian = "Exponential Gaussian"
+    SkewedGaussian = "Skewed Gaussian"
+    SkewedVoigt = "Skewed Voigt"
+    ThermalDistribution = "Thermal Distribution"
+    Doniach = "Doniach"
+    PowerLaw = "Power Law"
+    Exponential = "Exponential"
+    Step = "Step"
+    Rectangle = "Rectangle"
+    Expression = "Expression"
+    Constant = "Constant"
+
+    @classmethod
+    def default(cls):
+        return cls.Gaussian.value
+
+
+@NodeDecorator(id="span.basics.fit", name="Fit 1D")
+def fit_1D(
+    x_array: np.ndarray,
+    y_array: np.ndarray,
+    basic_peaks: List[PeakProperties],
+    model: FittingModel = FittingModel.default(),
+) -> List[PeakProperties]:
+    # """
+    # Fit a 1D model to the given data.
+
+    # Parameters:
+    #     peaks: dict
+    #         Dictionary containing the data and information about the peaks.
+    #     fitting_model: Optional[str]
+    #         The model to use for fitting. Defaults to "Gaussian".
+    #     preview: bool
+    #         Whether to preview the fit with a plot. Defaults to True.
+    #     color: Optional[Tuple[int, int, int] | str]
+    #         Color for the plot.
+
+    # Returns:
+    #     Tuple[dict, Optional[Figure]]:
+    #         A tuple containing a dictionary of evaluated components of the fit and additional information about the fit, and an optional figure for the plot.
+
+    # """
+    if isinstance(model, FittingModel):
+        model = model.value
+    peaks = copy.deepcopy(basic_peaks)
+    y = y_array
+    x = x_array
+    # if is_sturated:
+    #     if len(peaks['peaks']) == 2:
+    #         peaks['peaks'] = [{
+    #             "Peak #": 'peak1_',
+    #             "Index": peaks['peaks'][0]['Ending index'] + int( (peaks['peaks'][1]['Initial index'] - peaks['peaks'][0]['Ending index'])/ 2),
+    #             "Initial index": peaks['peaks'][0]['Ending index'],
+    #             "Ending index": peaks['peaks'][1]['Initial index'],
+    #             "Retention": np.NaN,
+    #             "Area": np.NaN,
+    #             "Height": y[peaks['peaks'][0]['Ending index'] + int( (peaks['peaks'][1]['Initial index'] - peaks['peaks'][0]['Ending index'])/ 2)],
+    #             "Symmetricity": np.NaN,
+    #             "Tailing": np.NaN,
+    #             "FWHM": np.NaN,
+    #             "Plate #": np.NaN,
+    #             "Width":  x[peaks['peaks'][1]['Initial index']] - x[peaks['peaks'][0]['Ending index']],
+    #             "is_fitted": False,
+    #         }]
+    #                     not_saturated_x = np.concatenate((x[:peaks['peaks'][0]['Ending index']-1], x[peaks['peaks'][1]['Initial index']+1:]))
+    #         not_saturated_y = np.concatenate((y[:peaks['peaks'][0]['Ending index']-1], y[peaks['peaks'][1]['Initial index']+1:]))
+    #     else:
+    #         raise ValueError('invalid number of peak selection. Either the entire or two sides of the saturated peak should be selected')
+
+    lowest_index = min(dictionary["i_index"] for dictionary in peaks)
+    highest_index = max(dictionary["f_index"] for dictionary in peaks)
+
+    # list of modelnames:
+    # lmfit.models.__dict__['lmfit_models'].keys()
+    # ['Constant', 'Complex Constant', 'Linear', 'Quadratic', 'Polynomial', 'Spline', 'Gaussian', 'Gaussian-2D', 'Lorentzian', 'Split-Lorentzian', 'Voigt', 'PseudoVoigt', 'Moffat', 'Pearson4', 'Pearson7', 'StudentsT', 'Breit-Wigner', 'Log-Normal', 'Damped Oscillator', 'Damped Harmonic Oscillator', 'Exponential Gaussian', 'Skewed Gaussian', 'Skewed Voigt', 'Thermal Distribution', 'Doniach', 'Power Law', 'Exponential', 'Step', 'Rectangle', 'Expression']
+    # peak like models are:  GaussianModel, LorentzianModel, VoigtModel and their modified versions
+
+    fitting_model = lmfit.models.__dict__["lmfit_models"][model]
+    # bkg1 = lmfit.models.__dict__["lmfit_models"]["Spline"](prefix="baseline", xknots=np.concatenate((x[:lowest_index], x[highest_index:])))
+    bkg2 = lmfit.models.__dict__["lmfit_models"]["Exponential"](prefix="baseline")
+
+    f = bkg2
+
+    pars = f.guess(y, x=x)
+    for index, peak in enumerate(peaks):
+        model = fitting_model(prefix=f"peak{index+1}_")
+        pars.update(model.make_params())
+        pars[f"peak{index+1}_center"].set(
+            value=x[peak["index"]],
+            min=x[peak["i_index"]],
+            max=x[peak["f_index"]],
+        )
+        pars[f"peak{index+1}_sigma"].set(
+            value=(x[peak["f_index"]] - x[peak["i_index"]]) / 2
+        )
+        pars[f"peak{index+1}_amplitude"].set(value=y[peak["index"]], min=0)
+
+        if model == "Exponential Gaussian" or model == "Skewed Gaussian":
+            pars[f"peak{index+1}_gamma"].set(value=1)
+
+        f += model
+
+    out = f.fit(y, pars, x=x)
+
+    f = bkg2
+    pars = f.guess(y, x=x)
+    for index, peak in enumerate(peaks):
+        model = fitting_model(prefix=f"peak{index+1}_")
+        pars.update(model.make_params())
+        pars[f"peak{index+1}_center"].set(
+            value=out.__dict__["best_values"][f"peak{index+1}_center"],
+            min=x[peak["i_index"]],
+            max=x[peak["f_index"]],
+        )
+        pars[f"peak{index+1}_sigma"].set(
+            value=(x[peak["f_index"]] - x[peak["i_index"]]) / 2
+        )
+        pars[f"peak{index+1}_amplitude"].set(
+            value=out.__dict__["best_values"][f"peak{index+1}_amplitude"], min=0
+        )
+
+        if model == "Exponential Gaussian" or model == "Skewed Gaussian":
+            pars[f"peak{index+1}_gamma"].set(
+                value=out.__dict__["best_values"][f"peak{index+1}_gamma"]
+            )
+
+        f += model
+
+    out = f.fit(y, pars, x=x)
+    com = out.eval_components(x=x)
+    info_dict = out.__dict__
+    info_dict["model_name"] = model
+
+    peak_properties_list = []
+
+    for key in com.keys():
+        if key != "baseline":
+            y_array = com[key]
+            peak_lst = [(0, np.argmax(y_array), len(y_array) - 1)]
+            for peak_nr, peak in enumerate(peak_lst):
+                peak_properties = compute_peak_properties(
+                    x_array=x_array,
+                    y_array=y_array,
+                    peak_indices=peak,
+                    peak_nr=peak_nr,
+                    is_fitted=True,
+                    fitting_data=com,
+                    fitting_info=info_dict,
+                )
+                peak_properties_list.append(peak_properties)
+
+    return peak_properties_list
+
+
+# Define a mapping from "C0", "C1", etc., to CSS color names
+color_map = {
+    "C0": "blue",
+    "C1": "orange",
+    "C2": "green",
+    "C3": "red",
+    "C4": "purple",
+    "C5": "brown",
+    "C6": "pink",
+    "C7": "gray",
+    "C8": "olive",
+    "C9": "cyan",
+}
+
+
+@NodeDecorator(id="span.basics.fit.plot", name="Plot fit 1D")
+def plot_fitted_peaks(peaks: List[PeakProperties]) -> go.Figure:
+    peak = peaks[0]
+    x = peak["fitting_info"]["userkws"]["x"]
+    # Extract data from peaks
+    y = peak["fitting_info"]["data"]
+    best_fit = peak["fitting_info"]["best_fit"]
+
+    # Create a subplot with 1 row, 1 column, and a secondary y-axis
+    fig = make_subplots(specs=[[{"secondary_y": True}]])
+
+    # Add the original data trace
+    fig.add_trace(
+        go.Scatter(
+            x=x, y=y, mode="lines", name="original", line=dict(color=color_map["C0"])
+        ),
+        secondary_y=False,
+    )
+
+    # Add the best fit trace
+    fig.add_trace(
+        go.Scatter(
+            x=x,
+            y=best_fit,
+            mode="lines",
+            name="best_fit",
+            line=dict(dash="dash", color=color_map["C1"]),
+        ),
+        secondary_y=False,
+    )
+
+    # Add the baseline and individual peak traces
+    for key in peak["fitting_data"].keys():
+        if key == "baseline":
+            color = color_map["C2"]
+        else:
+            peak_number = int(re.search(r"\d+", key).group())
+            color = color_map.get(
+                f"C{peak_number + 2}", "black"
+            )  # Default to black if not found
+
+        trace = go.Scatter(
+            x=x,
+            y=peak["fitting_data"][key],
+            mode="lines",
+            name=key,
+            line=dict(color=color),
+        )
+        fig.add_trace(trace, secondary_y=(key != "baseline"))
+
+    # Update axes labels and legend
+    fig.update_yaxes(title_text="Original", secondary_y=False)
+    fig.update_yaxes(title_text="Baseline corrected", secondary_y=True)
+    fig.update_layout(
+        title={
+            "text": f"{peak['fitting_info']['model_name']} model with fitting score = {np.round(peak['fitting_info']['rsquared'], 4)}",
+            "x": 0.5,  # Center the title
+            "xanchor": "center",
+        },
+    )
+
+    return fig
+
+
+PEAKS_NODE_SHELF = Shelf(
+    nodes=[peak_finder, fit_1D, plot_fitted_peaks],
+    subshelves=[],
+    name="Peak analysis",
+    description="Tools for the peak analysis of the spectra",
+)

{funcnodes_span-0.1.2 → funcnodes_span-0.1.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "funcnodes-span"
-version = "0.1.2"
+version = "0.1.4"
 description = ""
 authors = ["Kourosh Rezaei <kouroshrezaei90@gmail.com>"]
 readme = "README.md"
@@ -8,14 +8,19 @@ readme = "README.md"
 [tool.poetry.dependencies]
 python = "^3.11"
 scipy = "^1.14.0"
+
+lmfit = "^1.3.2"
 funcnodes = "^0.2.21"
 funcnodes_numpy = "^0.1.57"
 funcnodes_pandas = "^0.1.9"
+
+funcnodes_plotly = "^0.1.7"
+
 pooch = "^1.8.2"
 
 
 [tool.poetry.group.dev.dependencies]
-pytest = "^8.2.2"
+pytest = "^8.3.2"
 
 [build-system]
 requires = ["poetry-core"]