rapidtide 3.0a12__py3-none-any.whl → 3.0a14__py3-none-any.whl

rapidtide/fit.py

@@ -23,6 +23,7 @@ import matplotlib.pyplot as plt
 import numpy as np
 import scipy as sp
 import scipy.special as sps
+import statsmodels.api as sm
 import tqdm
 from numpy.polynomial import Polynomial
 from scipy.optimize import curve_fit
@@ -1105,6 +1106,37 @@ def mlproject(thefit, theevs, intercept):
     return thedest
 
 
+def olsregress(X, y, intercept=True, debug=False):
+    """
+
+    Parameters
+    ----------
+    X
+    y
+    intercept
+
+    Returns
+    -------
+
+    """
+    """Return the coefficients from a multiple linear regression, along with R, the coefficient of determination.
+
+    X: The independent variables (nxp).
+    y: The dependent variable (1xn or nx1).
+    intercept: Specifies whether or not the slope intercept should be considered.
+
+    The routine computes the coefficients (b_0, b_1, ..., b_p) from the data (x,y) under
+    the assumption that y = b0 + b_1 * x_1 + b_2 * x_2 + ... + b_p * x_p.
+
+    If intercept is False, the routine assumes that b0 = 0 and returns (b_1, b_2, ..., b_p).
+    """
+    if intercept:
+        X = sm.add_constant(X, prepend=True)
+    model = sm.OLS(y, exog=X)
+    thefit = model.fit()
+    return thefit.params, np.sqrt(thefit.rsquared)
+
+
 def mlregress(X, y, intercept=True, debug=False):
     """
 
@@ -1223,9 +1255,9 @@ def calcexpandedregressors(
     return outputregressors, outlabels
 
 
-def derivativeglmfilt(thedata, theevs, nderivs=1, debug=False):
+def derivativelinfitfilt(thedata, theevs, nderivs=1, debug=False):
     r"""First perform multicomponent expansion on theevs (each ev replaced by itself,
-    its square, its cube, etc.).  Then perform a glm fit of thedata using the vectors
+    its square, its cube, etc.).  Then perform a linear fit of thedata using the vectors
     in thenewevs and return the result.
 
     Parameters
@@ -1269,16 +1301,16 @@ def derivativeglmfilt(thedata, theevs, nderivs=1, debug=False):
     if debug:
         print(f"{nderivs=}")
         print(f"{thenewevs.shape=}")
-    filtered, datatoremove, R, coffs = glmfilt(thedata, thenewevs, debug=debug)
+    filtered, datatoremove, R, coffs = linfitfilt(thedata, thenewevs, debug=debug)
     if debug:
         print(f"{R=}")
 
     return filtered, thenewevs, datatoremove, R, coffs
 
 
-def expandedglmfilt(thedata, theevs, ncomps=1, debug=False):
+def expandedlinfitfilt(thedata, theevs, ncomps=1, debug=False):
     r"""First perform multicomponent expansion on theevs (each ev replaced by itself,
-    its square, its cube, etc.).  Then perform a glm fit of thedata using the vectors
+    its square, its cube, etc.).  Then perform a multiple regression fit of thedata using the vectors
     in thenewevs and return the result.
 
     Parameters
@@ -1322,15 +1354,15 @@ def expandedglmfilt(thedata, theevs, ncomps=1, debug=False):
     if debug:
         print(f"{ncomps=}")
         print(f"{thenewevs.shape=}")
-    filtered, datatoremove, R, coffs = glmfilt(thedata, thenewevs, debug=debug)
+    filtered, datatoremove, R, coffs = linfitfilt(thedata, thenewevs, debug=debug)
     if debug:
         print(f"{R=}")
 
     return filtered, thenewevs, datatoremove, R, coffs
 
 
-def glmfilt(thedata, theevs, returnintercept=False, debug=False):
-    r"""Performs a glm fit of thedata using the vectors in theevs
+def linfitfilt(thedata, theevs, returnintercept=False, debug=False):
+    r"""Performs a multiple regression fit of thedata using the vectors in theevs
     and returns the result.
 
     Parameters
@@ -1383,7 +1415,7 @@ def glmfilt(thedata, theevs, returnintercept=False, debug=False):
         return filtered, datatoremove, R2, retcoffs
 
 
-def confoundglm(
+def confoundregress(
     data,
     regressors,
     debug=False,

rapidtide/happy_supportfuncs.py

@@ -290,6 +290,11 @@ def cardiacfromimage(
     )
 
 
+def theCOM(X, data):
+    # return the center of mass
+    return np.sum(X * data) / np.sum(data)
+
+
 def savgolsmooth(data, smoothlen=101, polyorder=3):
     return savgol_filter(data, smoothlen, polyorder)
 

rapidtide/io.py

@@ -1426,6 +1426,8 @@ def writebidstsv(
     extraheaderinfo=None,
     compressed=True,
     columns=None,
+    xaxislabel="time",
+    yaxislabel="arbitrary value",
     starttime=0.0,
     append=False,
     colsinjson=True,
@@ -1447,6 +1449,8 @@ def writebidstsv(
     :param samplerate:
     :param compressed:
     :param columns:
+    :param xaxislabel:
+    :param yaxislabel:
     :param starttime:
     :param append:
     :param colsinjson:
@@ -1462,6 +1466,8 @@ def writebidstsv(
         print("\tsamplerate:", samplerate)
         print("\tcompressed:", compressed)
         print("\tcolumns:", columns)
+        print("\txaxislabel:", xaxislabel)
+        print("\tyaxislabel:", yaxislabel)
         print("\tstarttime:", starttime)
         print("\tappend:", append)
     if len(data.shape) == 1:
@@ -1538,6 +1544,8 @@ def writebidstsv(
     headerdict = {}
     headerdict["SamplingFrequency"] = float(samplerate)
     headerdict["StartTime"] = float(starttime)
+    headerdict["XAxisLabel"] = xaxislabel
+    headerdict["YAxisLabel"] = yaxislabel
     if colsinjson:
         if startcol == 0:
             headerdict["Columns"] = columns
@@ -1991,8 +1999,11 @@ def colspectolist(colspec, debug=False):
         ("APARC_SUBCORTGRAY", "8-13,17-20,26-28,47-56,58-60,96,97"),
         ("APARC_CORTGRAY", "1000-1035,2000-2035"),
         ("APARC_GRAY", "8-13,17-20,26-28,47-56,58-60,96,97,1000-1035,2000-2035"),
-        ("APARC_WHITE", "2,7,41,46,177,219"),
-        ("APARC_ALLBUTCSF", "2,7-13,17-20,26-28,41,46-56,58-60,96,97,177,219,1000-1035,2000-2035"),
+        ("APARC_WHITE", "2,7,41,46,177,219,3000-3035,4000-4035,5001,5002"),
+        (
+            "APARC_ALLBUTCSF",
+            "2,7-13,17-20,26-28,41,46-56,58-60,96,97,177,219,1000-1035,2000-2035,3000-3035,4000-4035,5001,5002",
+        ),
         ("SSEG_GRAY", "3,8,10-13,16-18,26,42,47,49-54,58"),
         ("SSEG_WHITE", "2,7,41,46"),
     )

rapidtide/{glmpass.py → linfitfiltpass.py}

@@ -26,7 +26,9 @@ import rapidtide.miscmath as tide_math
 import rapidtide.multiproc as tide_multiproc
 
 
-def _procOneGLMItem(vox, theevs, thedata, rt_floatset=np.float64, rt_floattype="float64"):
+def _procOneRegressionFitItem(
+    vox, theevs, thedata, rt_floatset=np.float64, rt_floattype="float64"
+):
     # NOTE: if theevs is 2D, dimension 0 is number of points, dimension 1 is number of evs
     thefit, R2 = tide_fit.mlregress(theevs, thedata)
     if theevs.ndim > 1:
@@ -75,7 +77,7 @@ def _procOneGLMItem(vox, theevs, thedata, rt_floatset=np.float64, rt_floattype="
         )
 
 
-def glmpass(
+def linfitfiltpass(
     numprocitems,
     fmri_data,
     threshval,
@@ -89,12 +91,13 @@ def glmpass(
     filtereddata,
     nprocs=1,
     alwaysmultiproc=False,
-    confoundglm=False,
+    confoundregress=False,
     procbyvoxel=True,
     showprogressbar=True,
     mp_chunksize=1000,
     rt_floatset=np.float64,
     rt_floattype="float64",
+    verbose=True,
     debug=False,
 ):
     inputshape = np.shape(fmri_data)
@@ -132,9 +135,9 @@ def glmpass(
 
                     # process and send the data
                     if procbyvoxel:
-                        if confoundglm:
+                        if confoundregress:
                             outQ.put(
-                                _procOneGLMItem(
+                                _procOneRegressionFitItem(
                                     val,
                                     theevs,
                                     fmri_data[val, :],
@@ -144,7 +147,7 @@ def glmpass(
                             )
                         else:
                             outQ.put(
-                                _procOneGLMItem(
+                                _procOneRegressionFitItem(
                                     val,
                                     theevs[val, :],
                                     fmri_data[val, :],
@@ -153,9 +156,9 @@ def glmpass(
                                 )
                             )
                     else:
-                        if confoundglm:
+                        if confoundregress:
                             outQ.put(
-                                _procOneGLMItem(
+                                _procOneRegressionFitItem(
                                     val,
                                     theevs,
                                     fmri_data[:, val],
@@ -165,7 +168,7 @@ def glmpass(
                             )
                         else:
                             outQ.put(
-                                _procOneGLMItem(
+                                _procOneRegressionFitItem(
                                     val,
                                     theevs[:, val],
                                     fmri_data[:, val],
@@ -182,6 +185,7 @@ def glmpass(
             GLM_consumer,
             inputshape,
             themask,
+            verbose=verbose,
             nprocs=nprocs,
             procbyvoxel=procbyvoxel,
             showprogressbar=showprogressbar,
@@ -191,7 +195,7 @@ def glmpass(
         # unpack the data
         itemstotal = 0
         if procbyvoxel:
-            if confoundglm:
+            if confoundregress:
                 for voxel in data_out:
                     r2value[voxel[0]] = voxel[3]
                     filtereddata[voxel[0], :] = voxel[7]
@@ -211,7 +215,7 @@ def glmpass(
                     filtereddata[voxel[0], :] = voxel[7]
                     itemstotal += 1
         else:
-            if confoundglm:
+            if confoundregress:
                 for timepoint in data_out:
                     r2value[timepoint[0]] = timepoint[3]
                     filtereddata[:, timepoint[0]] = timepoint[7]
@@ -243,7 +247,7 @@ def glmpass(
             ):
                 thedata = fmri_data[vox, :].copy()
                 if (themask is None) or (themask[vox] > 0):
-                    if confoundglm:
+                    if confoundregress:
                         (
                             dummy,
                             dummy,
@@ -253,7 +257,7 @@ def glmpass(
                             dummy,
                             dummy,
                             filtereddata[vox, :],
-                        ) = _procOneGLMItem(
+                        ) = _procOneRegressionFitItem(
                             vox,
                             theevs,
                             thedata,
@@ -270,7 +274,7 @@ def glmpass(
                             fitNorm[vox],
                             datatoremove[vox, :],
                             filtereddata[vox, :],
-                        ) = _procOneGLMItem(
+                        ) = _procOneRegressionFitItem(
                             vox,
                             theevs[vox, :],
                             thedata,
@@ -287,7 +291,7 @@ def glmpass(
             ):
                 thedata = fmri_data[:, timepoint].copy()
                 if (themask is None) or (themask[timepoint] > 0):
-                    if confoundglm:
+                    if confoundregress:
                         (
                             dummy,
                             dummy,
@@ -297,7 +301,7 @@ def glmpass(
                             dummy,
                             dummy,
                             filtereddata[:, timepoint],
-                        ) = _procOneGLMItem(
+                        ) = _procOneRegressionFitItem(
                             timepoint,
                             theevs,
                             thedata,
@@ -314,7 +318,7 @@ def glmpass(
                             fitNorm[timepoint],
                             datatoremove[:, timepoint],
                             filtereddata[:, timepoint],
-                        ) = _procOneGLMItem(
+                        ) = _procOneRegressionFitItem(
                             timepoint,
                             theevs[:, timepoint],
                             thedata,
@@ -425,7 +429,7 @@ def confoundregress(
     numprocitems = thedataarray.shape[0]
     filtereddata = thedataarray * 0.0
     r2value = np.zeros(numprocitems)
-    numfiltered = glmpass(
+    numfiltered = linfitfiltpass(
         numprocitems,
         thedataarray,
         None,
@@ -437,7 +441,7 @@ def confoundregress(
         None,
         None,
         filtereddata,
-        confoundglm=True,
+        confoundregress=True,
         nprocs=nprocs,
         showprogressbar=showprogressbar,
         procbyvoxel=True,

rapidtide/makelaggedtcs.py

@@ -24,8 +24,6 @@ from tqdm import tqdm
 
 import rapidtide.multiproc as tide_multiproc
 
-LGR = logging.getLogger("GENERAL")
-
 
 def _procOneVoxelMakelagtc(
     vox,
@@ -51,6 +49,7 @@ def makelaggedtcs(
     lagmask,
     lagtimes,
     lagtc,
+    LGR=None,
     nprocs=1,
     alwaysmultiproc=False,
     showprogressbar=True,
@@ -90,6 +89,7 @@ def makelaggedtcs(
             makelagtc_consumer,
             inputshape,
             lagmask,
+            verbose=(LGR is not None),
             nprocs=nprocs,
             showprogressbar=showprogressbar,
             chunksize=chunksize,
@@ -127,13 +127,16 @@ def makelaggedtcs(
                 )
                 volumetotal += 1
 
-    LGR.info(f"\nLagged timecourses created for {volumetotal} voxels")
+    if LGR is not None:
+        LGR.info(f"\nLagged timecourses created for {volumetotal} voxels")
 
     # garbage collect
     uncollected = gc.collect()
     if uncollected != 0:
-        LGR.info(f"garbage collected - unable to collect {uncollected} objects")
+        if LGR is not None:
+            LGR.info(f"garbage collected - unable to collect {uncollected} objects")
     else:
-        LGR.info("garbage collected")
+        if LGR is not None:
+            LGR.info("garbage collected")
 
     return volumetotal

rapidtide/multiproc.py

@@ -77,8 +77,8 @@ def _process_data(data_in, inQ, outQ, showprogressbar=True, reportstep=1000, chu
                 pbar.update(1)
                 if numreturned > remainder - 1:
                     break
-
-    print()
+    if showprogressbar:
+        print()
 
     return data_out
 
@@ -88,6 +88,7 @@ def run_multiproc(
     inputshape,
     maskarray,
     nprocs=1,
+    verbose=True,
     procbyvoxel=True,
     showprogressbar=True,
     chunksize=1000,
@@ -105,13 +106,6 @@ def run_multiproc(
         workers = [ctx.Process(target=consumerfunc, args=(inQ, outQ)) for i in range(n_workers)]
         # signal.signal(signal.SIGINT, original_sigint_handler)
     else:
-        """# try adding this magic incantation to get coverage to record multiprocessing properly
-        # This fails for python 3.8 and above
-        try:
-            from pytest_cov.embed import cleanup
-        except ImportError:
-            cleanup = None
-        """
         cleanup = None  # just disable this for now
         inQ = mp.Queue()
         outQ = mp.Queue()
@@ -141,7 +135,8 @@ def run_multiproc(
             data_in.append(d)
         elif maskarray[d] > 0.5:
             data_in.append(d)
-    print("processing", len(data_in), procunit + " with", n_workers, "processes")
+    if verbose:
+        print("processing", len(data_in), procunit + " with", n_workers, "processes")
     data_out = _process_data(
         data_in, inQ, outQ, showprogressbar=showprogressbar, chunksize=chunksize
     )
@@ -162,6 +157,7 @@ def run_multithread(
     consumerfunc,
     inputshape,
     maskarray,
+    verbose=True,
     nprocs=1,
     procbyvoxel=True,
     showprogressbar=True,
@@ -197,7 +193,8 @@ def run_multithread(
             data_in.append(d)
         elif maskarray[d] > 0:
             data_in.append(d)
-    print("processing", len(data_in), procunit + " with", n_workers, "threads")
+    if verbose:
+        print("processing", len(data_in), procunit + " with", n_workers, "threads")
     data_out = _process_data(
         data_in, inQ, outQ, showprogressbar=showprogressbar, chunksize=chunksize
     )