Build is running again

.github/workflows/build.yml

@@ -8,7 +8,7 @@ jobs:
     steps:
     - uses: actions/checkout@v3
     - name: Install dependencies
-      run: poetry install --no-root
+      run: poetry install
     - name: Build figures
       run: make all -j 3
     - name: Upload files

.github/workflows/test.yml

@@ -8,6 +8,6 @@ jobs:
     steps:
     - uses: actions/checkout@v3
     - name: Install dependencies
-      run: poetry install --no-root
+      run: poetry install
     - name: Test with pytest
       run: poetry run pytest --junitxml=junit.xml --cov-branch --cov=ckine --cov-report xml:coverage.xml

ckine/FCimports.py

@@ -6,7 +6,7 @@
 from pathlib import Path
 import pandas as pd
 import numpy as np
-from FlowCytometryTools import FCMeasurement, PolyGate
+from FlowCytometryTools import FCMeasurement
 
 path_here = dirname(dirname(__file__))
 

ckine/MBmodel.py

@@ -78,7 +78,7 @@ def cytBindingModel(mut, val, doseVec, cellType, x=False, date=False):
 
     mutAffDF = pd.read_csv(join(path_here, "ckine/data/WTmutAffData.csv"))
     Affs = mutAffDF.loc[(mutAffDF.Mutein == mut)]
-    Affs = np.power(np.array([Affs["IL2RaKD"].values, Affs["IL2RBGKD"].values]) / 1e9, -1)
+    Affs = np.power(np.array([Affs["IL2RaKD"].values, Affs["IL2RBGKD"].values], dtype=float) / 1e9, -1)
     Affs = np.reshape(Affs, (1, -1))
     Affs = np.repeat(Affs, 2, axis=0)
     np.fill_diagonal(Affs, 1e2)  # Each cytokine can only bind one a and one b
@@ -109,8 +109,8 @@ def runFullModel(x=False, time=[0.5], saveDict=False, singleCell=False):
     statDF = statDF.loc[(statDF.Ligand != "H16L N-term (Mono)") & (statDF.Ligand != "IL15 (Mono)")]
     statDF = statDF.loc[(statDF.Time.isin(time))]
 
-    dateConvDF = pd.DataFrame(columns={"Date", "Scale", "Cell"})
-    masterSTAT = pd.DataFrame(columns={"Ligand", "Date", "Cell", "Time", "Dose", "Valency", "Experimental", "Predicted"})
+    dateConvDF = pd.DataFrame(columns=["Date", "Scale", "Cell"])
+    masterSTAT = pd.DataFrame(columns=["Ligand", "Date", "Cell", "Time", "Dose", "Valency", "Experimental", "Predicted"])
     dates = statDF.Date.unique()
 
     for (date, lig, conc, cell, time), group in statDF.groupby(["Date", "Ligand", "Dose", "Cell", "Time"]):
@@ -156,8 +156,8 @@ def runFullModelMeyer(x=False, saveDict=False):
     """Runs model for all data points and outputs date conversion dict for binding to pSTAT. Can be used to fit Kx"""
     statDF = import_pstat_all_meyer()
 
-    dateConvDF = pd.DataFrame(columns={"Date", "Scale", "Cell"})
-    masterSTAT = pd.DataFrame(columns={"Ligand", "Date", "Cell", "Dose", "Valency", "Experimental", "Predicted"})
+    dateConvDF = pd.DataFrame(columns=["Date", "Scale", "Cell"])
+    masterSTAT = pd.DataFrame(columns=["Ligand", "Date", "Cell", "Dose", "Valency", "Experimental", "Predicted"])
     dates = statDF.Date.unique()
 
     for (date, lig, val, conc, cell), group in statDF.groupby(["Date", "Ligand", "Valency", "Dose", "Cell"]):

ckine/PCA.py

@@ -419,7 +419,7 @@ def StatGini(sampleType, ax, cell_type, date, Tcells=True):
 
 def nllsq_EC50(x0, xdata, ydata):
     """ Performs nonlinear least squares on activity measurements to determine parameters of Hill equation and outputs EC50. """
-    lsq_res = least_squares(residuals, x0, args=(xdata, ydata), bounds=([0.0, 0.0, 0.0], [10, 10.0, 10 ** 5.0]), jac="3-point")
+    lsq_res = least_squares(residuals, x0, method="dogbox", args=(xdata, ydata), bounds=([0.0, 0.0, 0.0], [10, 10.0, 10 ** 5.0]), jac="3-point")
     return lsq_res.x[0]
 
 

ckine/figures/figureC1.py

@@ -110,7 +110,7 @@ def fullHeatMap(ax, respDF):
 def dosePlot(ax, respDF, time, cell, ligList=False, legend=False):
     """Plots the various affinities for IL-2 Muteins"""
     doses = np.log10(np.logspace(np.log10(respDF.Dose.min()), np.log10(respDF.Dose.max()), 100)) + 4
-    x0 = [4, 1, 2]
+    x0 = [4.0, 1.0, 2.0]
     hillDF = pd.DataFrame()
     if not ligList:
         Ligands = respDF.Ligand.unique()
@@ -125,7 +125,7 @@ def dosePlot(ax, respDF, time, cell, ligList=False, legend=False):
             isoData = respDF.loc[(respDF.Ligand == ligand) & (respDF.Valency == valency)]
             xData = np.nan_to_num(np.log10(isoData.Dose.values)) + 4
             yData = np.nan_to_num(isoData.Mean.values)
-            fit = least_squares(hill_residuals, x0, args=(xData, yData), bounds=([0.0, 0.0, 2], [5, 10.0, 6]), jac="3-point")
+            fit = least_squares(hill_residuals, x0, method="dogbox", args=(xData, yData), bounds=([0.0, 0.0, 2], [5, 10.0, 6]), jac="3-point")
             hillDF = pd.concat([hillDF, pd.DataFrame({"Ligand": ligand, "Valency": valency, "Dose": np.power(10, doses - 4), "pSTAT": hill_equation(fit.x, doses)})])
 
     maxobs = hillDF.loc[(hillDF.Ligand == "IL2")].pSTAT.max()

ckine/figures/figureC2.py

@@ -7,10 +7,9 @@
 import seaborn as sns
 import numpy as np
 import matplotlib.pyplot as plt
-from scipy.optimize import least_squares
 from os.path import join
 from copy import copy
-from .figureCommon import subplotLabel, getSetup, getLigDict, get_cellTypeDict, getLigandLegend, Wass_KL_Dist, CITE_RIDGE, hillRatioDosePlot
+from .figureCommon import subplotLabel, getSetup, getLigDict, get_cellTypeDict, getLigandLegend, Wass_KL_Dist, hillRatioDosePlot
 from ..imports import import_pstat_all
 
 path_here = os.path.dirname(os.path.dirname(__file__))

ckine/figures/figureC4.py

@@ -8,7 +8,6 @@
 import seaborn as sns
 import matplotlib.pyplot as plt
 from sklearn.metrics import r2_score
-from scipy.optimize import minimize
 from copy import copy
 from .figureCommon import subplotLabel, getSetup, get_cellTypeDict, get_doseLimDict, get_cellTypeDict, get_valency_dict, ligand_ratio_plot
 from ..PCA import nllsq_EC50
@@ -67,7 +66,7 @@ def Pred_Exp_plot(ax, df):
 
 def R2_Plot_Cells(ax, df):
     """Plots all accuracies per cell"""
-    accDF = pd.DataFrame(columns={"Cell Type", "Valency", "Accuracy"})
+    accDF = pd.DataFrame(columns=["Cell Type", "Valency", "Accuracy"])
     cellTypes = ['Treg $IL2Ra^{hi}$', 'Treg', 'Treg $IL2Ra^{lo}$', 'Thelper $IL2Ra^{hi}$', 'Thelper', 'Thelper $IL2Ra^{lo}$', 'CD8', 'NK', 'NKBright']
     for cell in cellTypes:
         for val in df.Valency.unique():
@@ -84,7 +83,7 @@ def R2_Plot_Cells(ax, df):
 
 def R2_Plot_Ligs(ax, df):
     """Plots all accuracies per ligand"""
-    accDF = pd.DataFrame(columns={"Ligand", "Valency", "Accuracy"})
+    accDF = pd.DataFrame(columns=["Ligand", "Valency", "Accuracy"])
     for ligand in df.Ligand.unique():
         for val in df.loc[df.Ligand == ligand].Valency.unique():
             preds = df.loc[(df.Ligand == ligand) & (df.Valency == val)].Predicted.values
@@ -100,7 +99,7 @@ def R2_Plot_Ligs(ax, df):
 
 def R2_Plot_Conc(ax, df):
     """Plots all accuracies per concentration"""
-    accDF = pd.DataFrame(columns={"Concentration", "Valency", "Accuracy"})
+    accDF = pd.DataFrame(columns=["Concentration", "Valency", "Accuracy"])
     for conc in df.Dose.unique():
         for val in df.loc[(df.Dose == conc)].Valency.unique():
             preds = df.loc[(df.Dose == conc) & (df.Valency == val)].Predicted.values
@@ -148,7 +147,7 @@ def MonVsBivalent(ax, dfAll, ligs=True):
                 df.loc[(df.Date == date) & (df.Cell == cell), "MonPredict"] = predVec * slope
 
     if ligs:
-        accDF = pd.DataFrame(columns={"Ligand", "Prediction Label", "Accuracy"})
+        accDF = pd.DataFrame(columns=["Ligand", "Prediction Label", "Accuracy"])
         for ligand in df.Ligand.unique():
             BivPreds = df.loc[(df.Ligand == ligand)].Predicted.values
             MonPreds = df.loc[(df.Ligand == ligand)].MonPredict.values
@@ -219,7 +218,7 @@ def EC50comp(ax, dfAll, time):
 def timePlot(ax):
     """Plots all experimental vs. Predicted Values"""
     times = [[0.5], [1.], [2.], [4.]]
-    accDF = pd.DataFrame(columns={"Time", "Valency", "Accuracy"})
+    accDF = pd.DataFrame(columns=["Time", "Valency", "Accuracy"])
     for time in times:
         df = runFullModel(time=time, saveDict=False)
         for val in df.Valency.unique():

ckine/figures/figureC5.py

@@ -12,7 +12,6 @@
 from .figureCommon import subplotLabel, getSetup, ligand_ratio_plot
 from ..MBmodel import polyc, getKxStar, runFullModelMeyer
 from ..imports import getBindDict, importReceptors
-from ..flow_meyer import make_flow_df, make_flow_df_ILCs
 
 path_here = dirname(dirname(__file__))
 plt.rcParams['svg.fonttype'] = 'none'
@@ -91,10 +90,10 @@ def optimizeDesign(ax, targCell, offTcells, IL7=False, legend=True):
     sigDF = pd.DataFrame()
 
     if IL7:
-        optDF = pd.DataFrame(columns={"Valency", "Selectivity", "IL7Rα"})
+        optDF = pd.DataFrame(columns=["Valency", "Selectivity", "IL7Rα"])
         X0 = [8]  # Ka IL7
     else:
-        optDF = pd.DataFrame(columns={"Valency", "Selectivity", "IL2Rα", r"IL-2Rβ/γ$_c$"})
+        optDF = pd.DataFrame(columns=["Valency", "Selectivity", "IL2Rα", r"IL-2Rβ/γ$_c$"])
         if targCell[0] == "NK":
             X0 = [6.0, 8]  # IL2Ra, IL2Rb
         else:

ckine/figures/figureCommon.py

@@ -466,12 +466,12 @@ def hillRatioDosePlot(ax, respDF, time, targCell, offTargCell, pseudo=0.2, plot=
             targIsoData = respDF.loc[(respDF.Ligand == ligand) & (respDF.Valency == valency) & (respDF.Cell == targCell)]
             targXData = np.nan_to_num(np.log10(targIsoData.Dose.values)) + 4
             targYData = np.nan_to_num(targIsoData.Mean.values)
-            targFit = least_squares(hill_residuals, x0, args=(targXData, targYData), bounds=([0.0, 0.0, 2], [5, 10.0, 6]), jac="3-point")
+            targFit = least_squares(hill_residuals, x0, method="dogbox", args=(targXData, targYData), bounds=([0.0, 0.0, 2], [5, 10.0, 6]), jac="3-point")
 
             offTIsoData = respDF.loc[(respDF.Ligand == ligand) & (respDF.Valency == valency) & (respDF.Cell == offTargCell)]
             offTargXData = np.nan_to_num(np.log10(offTIsoData.Dose.values)) + 4
             offTargYData = np.nan_to_num(offTIsoData.Mean.values)
-            offTargFit = least_squares(hill_residuals, x0, args=(offTargXData, offTargYData), bounds=([0.0, 0.0, 2], [5, 10.0, 6]), jac="3-point")
+            offTargFit = least_squares(hill_residuals, x0, method="dogbox", args=(offTargXData, offTargYData), bounds=([0.0, 0.0, 2], [5, 10.0, 6]), jac="3-point")
             hillDF = pd.concat([hillDF, pd.DataFrame({"Ligand": ligand, "Valency": valency, "Cell": targCell, "Dose": np.power(
                 10, doses - 4), targCell: hill_equation(targFit.x, doses), offTargCell: hill_equation(offTargFit.x, doses)})])
 
@@ -519,7 +519,7 @@ def make_EC50_DF(respDF, time, meyer=False):
                 targIsoData = respDF.loc[(respDF.Ligand == ligand) & (respDF.Valency == valency) & (respDF.Cell == cell)]
                 targXData = np.nan_to_num(np.log10(targIsoData.Dose.values)) + 4
                 targYData = np.nan_to_num(targIsoData.Mean.values)
-                targFit = least_squares(hill_residuals, x0, args=(targXData, targYData), bounds=([0.0, 0.0, 2], [5, 10.0, 6]), jac="3-point")
+                targFit = least_squares(hill_residuals, x0, method="dogbox", args=(targXData, targYData), bounds=([0.0, 0.0, 2], [5, 10.0, 6]), jac="3-point")
                 EC50_DF = pd.concat([EC50_DF, pd.DataFrame({"Cell": cell, "Ligand": ligand, "Valency": valency, "EC50": [np.power(10, targFit.x[2] - 4)]})])
 
     EC50_DF.loc[(EC50_DF.Valency == 1), "Ligand"] = (EC50_DF.loc[(EC50_DF.Valency == 1)].Ligand + " (Mono)").values

ckine/figures/figureS2.py

@@ -7,7 +7,6 @@
 import seaborn as sns
 import numpy as np
 import matplotlib.pyplot as plt
-from scipy.optimize import least_squares
 from os.path import join
 from copy import copy
 from .figureCommon import subplotLabel, getSetup, getLigDict, get_cellTypeDict, hillRatioDosePlot

ckine/figures/figureS3.py

@@ -3,13 +3,7 @@
 """
 
 import os
-import pandas as pd
-import seaborn as sns
-import numpy as np
 import matplotlib.pyplot as plt
-from scipy.optimize import least_squares
-from os.path import join
-from copy import copy
 from .figureCommon import subplotLabel, getSetup, getLigDict, get_cellTypeDict, CITE_RIDGE, CITE_SVM
 
 

ckine/figures/figureS8.py

@@ -4,10 +4,8 @@
 
 import os
 import pandas as pd
-import seaborn as sns
 import numpy as np
 import matplotlib.pyplot as plt
-from scipy.optimize import least_squares
 from os.path import join
 from copy import copy
 from .figureCommon import subplotLabel, getSetup, getLigDict, get_cellTypeDict, make_EC50_DF

ckine/flow.py

@@ -7,7 +7,7 @@
 import numpy as np
 import pandas as pd
 import warnings
-from FlowCytometryTools import FCMeasurement, QuadGate, ThresholdGate, PolyGate
+from FlowCytometryTools import FCMeasurement, QuadGate, ThresholdGate
 
 path_here = dirname(dirname(__file__))
 
@@ -175,7 +175,7 @@ def nllsq(x, y):
     upper = np.array([1.0, 1.1, 1.0e6, 1.0e9])
     x0 = (upper - lower) / 2.0 + lower
 
-    lsq = least_squares(lambda pp: exp_dec(x, pp) - y, x0, bounds=(lower, upper), jac="3-point")
+    lsq = least_squares(lambda pp: exp_dec(x, pp) - y, x0, method="dogbox", bounds=(lower, upper), jac="3-point")
     return lsq.x