NiPype: Error inputting to SpecifySPMModel

1

Dear All,

I have a preprocessing worklfow that appears to run without any problems. I then connect this to the 1st level workflow:

metaflow.connect([(preproc, l1analysis, [('realign.realignment_parameters',
                                      'modelspec.realignment_parameters'),
                                     ('smooth.smoothed_files',
                                      'modelspec.functional_runs'),
                                     ('art.outlier_files',
                                      'modelspec.outlier_files'),

The inputs are then connected to this:

metaflow.connect([(infosource, selectfiles, [('subject_id', 'subject_id'), ('session_id', 'session_id')]),
              (selectfiles, preproc, [('func', 'reset_origin_func.niftis')]),
              (selectfiles, preproc, [('struc', 'reset_origin_struc.niftis')]),
              (infosource, getsubjectinfo, [('subject_id', 'subject_id')]),
              (getsubjectinfo, l1analysis, [('subject_info', 'modelspec.subject_info')]),
              (infosource, l1analysis, [('contrasts', 'conestimate.contrasts')]),

It seems that it is at the connection between getsubejctinfo and l1analysis the pipeline stumbles: I receive an error message:

"TypeError: can't multiply sequence by non-int of type 'float'
Interface SpecifySPMModel failed to run."

The firstlevel node that it fails on is:

modelspec = Node(SpecifySPMModel(concatenate_runs=False,
                             input_units='secs', output_units='secs',
                             time_repetition=TR,
                             high_pass_filter_cutoff=128),
             name="modelspec")

I wondered if this may be related to me inputting the task details incorrectly - this is the getsubjectinfo function that connects to the modelspec node:

def get_subject_info(subject_id):
    from nipype.interfaces.base import Bunch
    condition_names = ['rest', 'control', 'stress']

    onsets_run1 = [[0, 220.5], [10.5, 231], [98.7, 319.2]]
    onsets_run3 = [[0], [10.5], [98.7]]
    onset_list = [(onsets_run1), (onsets_run1), (onsets_run3)]

    durations_run1 = [[10.5],[88.2],[98.7]]
    durations_run3 = [[10.5],[88.2],[342.3]]
    durations_list = [(durations_run1), (durations_run1), (durations_run3)]

    subjectinfo = []
    for r in range(1):
        subjectinfo.insert(r,
                       Bunch(conditions=condition_names,
                             onsets=[onset_list[r]],
                             durations=[durations_list[r]],
                             amplitudes=None,
                             tmod=None,
                             pmod=None,
                             regressor_names=None,
                             regressors=None))
return subjectinfo

getsubjectinfo = Node(Function(input_names=['subject_id'], output_names=['subject_info'],
                           function=get_subject_info),
                  name='getsubjectinfo')]

From the error message I wondered whether some timings were being inputted as a string but this doesn’t seem to be the case.

Any help most appreciated!

Many Thanks

Rob

2

Hi @rob, did a crashfile get produced? If so, that may be a better way to debug the problem

3

Thanks mgxd - the crash file is below, I wasn’t really able to decipher it though

I wondered if this line pointed to the problem:

timelist = [np.max([0., _scalefactor * t]) for t in timelist]

Suggesting there is a problem with my list of timings but I can’t tell what this problem may be

  File: crash-20170317-204114-k1201869-modelspec.a0-f6aac261-95f3-43dd-aa91-a8cfaf994a71.pklz
    Node: metaflow.l1analysis.modelspec.a0
    Working directory: /home/k1201869/restingstate2017/workingdir_prepro/metaflow/l1analysis/_session_id_run1_subject_id_subject001/modelspec

Node inputs:

concatenate_runs = False
event_files = <undefined>
functional_runs = ['/home/k1201869/restingstate2017/workingdir_prepro/metaflow/preproc/_session_id_run1_subject_id_subject001/smooth/swrrun1_reset.nii']
high_pass_filter_cutoff = 128.0
ignore_exception = False
input_units = secs
outlier_files = ['/home/k1201869/restingstate2017/workingdir_prepro/metaflow/preproc/_session_id_run1_subject_id_subject001/art/art.rrun1_reset_outliers.txt']
output_units = secs
realignment_parameters = ['/home/k1201869/restingstate2017/workingdir_prepro/metaflow/preproc/_session_id_run1_subject_id_subject001/realign/rp_run1_reset.txt']
subject_info = [Bunch(amplitudes=None, conditions=['rest', 'control', 'stress'], durations=[[[10.5], [88.2], [98.7]]], onsets=[[[0, 220.5], [10.5, 231], [98.7, 319.2]]], pmod=None, regressor_names=None, regressors=None, tmod=None), 
Bunch(amplitudes=None, conditions=['rest', 'control', 'stress'], durations=[[[10.5], [88.2], [98.7]]], onsets=[[[0, 220.5], [10.5, 231], [98.7, 319.2]]], pmod=None, regressor_names=None, regressors=None, tmod=None), 
Bunch(amplitudes=None, conditions=['rest', 'control', 'stress'], durations=[[[10.5], [88.2], [342.3]]], onsets=[[[0], [10.5], [98.7]]], pmod=None, regressor_names=None, regressors=None, tmod=None)]
time_repetition = 2.1


Traceback: 
Traceback (most recent call last):
  File "/home/k1201869/restingstate2017/workingdir_prepro/metaflow/batch/pyscript_20170317_204013_metaflow.l1analysis_modelspec.a0.py", line 32, in <module>
result = info['node'].run(updatehash=info['updatehash'])
  File "/home/k1201869/anaconda2/lib/python2.7/site-packages/nipype/pipeline/engine/nodes.py", line 394, in run
self._run_interface()
  File "/home/k1201869/anaconda2/lib/python2.7/site-packages/nipype/pipeline/engine/nodes.py", line 504, in _run_interface
self._result = self._run_command(execute)
  File "/home/k1201869/anaconda2/lib/python2.7/site-packages/nipype/pipeline/engine/nodes.py", line 630, in _run_command
result = self._interface.run()
  File "/home/k1201869/anaconda2/lib/python2.7/site-packages/nipype/interfaces/base.py", line 1043, in run
runtime = self._run_wrapper(runtime)
  File "/home/k1201869/anaconda2/lib/python2.7/site-packages/nipype/interfaces/base.py", line 1000, in _run_wrapper
runtime = self._run_interface(runtime)
  File "/home/k1201869/anaconda2/lib/python2.7/site-packages/nipype/algorithms/modelgen.py", line 408, in _run_interface
self._generate_design()
  File "/home/k1201869/anaconda2/lib/python2.7/site-packages/nipype/algorithms/modelgen.py", line 522, in _generate_design
super(SpecifySPMModel, self)._generate_design(infolist=infolist)
  File "/home/k1201869/anaconda2/lib/python2.7/site-packages/nipype/algorithms/modelgen.py", line 402, in _generate_design
outliers=outliers)
  File "/home/k1201869/anaconda2/lib/python2.7/site-packages/nipype/algorithms/modelgen.py", line 309, in _generate_standard_design
self.inputs.time_repetition)
  File "/home/k1201869/anaconda2/lib/python2.7/site-packages/nipype/algorithms/modelgen.py", line 147, in scale_timings
timelist = [np.max([0., _scalefactor * t]) for t in timelist]
TypeError: can't multiply sequence by non-int of type 'float'
Interface SpecifySPMModel failed to run.
4

The problem could be due to the number of lists you have in your subject_info.

  • onsets=[[[0, 220.5], [10.5, 231], [98.7, 319.2]]]

should probably be

  • onsets=[[0, 220.5], [10.5, 231], [98.7, 319.2]]

durations should also be structured like this. There a nice example in the docs found here

5

Thanks again - the pipeline runs all the way through now. My one final problem is that it generates contrasts separately for each individual run. I don not want to concatenate the runs, but I would like all runs to be modelled at once so I just have one set of contrasts per subject (i.e. as one would normally do in SPM).

I think the issue is again in the get_subject_info function:

# Contrasts
cont01 = ['stress>control', 'T', condition_names, [0, -1, 1]]
cont02 = ['control>stress', 'T', condition_names, [0, 1, -1]]
contrast_list = [cont01, cont02]

# Task onsets and durations
def get_subject_info(subject_id):
    from nipype.interfaces.base import Bunch
    condition_names = ['rest', 'control', 'stress']
    onset_run1_rest = [0, 220.5]
    onset_run1_control = [10.5, 231]
    onset_run1_stress = [98.7, 319.2]
    onset_run3_rest = [0]
    onset_run3_control = [10.5]
    onset_run3_stress = [98.7]

    onsets_run3 = [[0], [10.5], [98.7]]
    onset_list = [(onset_run1_rest), (onset_run1_control), (onset_run1_stress),
                  (onset_run1_rest), (onset_run1_control), (onset_run1_stress),
                  (onset_run3_rest), (onset_run3_control), (onset_run3_stress)]

    durations_run1 = [[10.5],[88.2],[98.7]]
    durations_run3 = [[10.5],[88.2],[342.3]]
    durations_list = [(durations_run1), (durations_run1), (durations_run3)]

    subjectinfo = []
    for r in range(1):
        onsets = [onset_list[r*2],onset_list[r*2+1], onset_list[r*2+2]]
        subjectinfo.insert(r,
                           Bunch(conditions=condition_names,
                                 onsets=onsets,
                                 durations=durations_list[r],
                                 amplitudes=None,
                                 tmod=None,
                                 pmod=None,
                                 regressor_names=None,
                                 regressors=None))
    return subjectinfo

getsubjectinfo = Node(Function(input_names=['subject_id'], output_names=['subject_info'],
                               function=get_subject_info),
                      name='getsubjectinfo')

There are 3 runs but if I put anything other than ‘1’ in:

for r in range(1):

the pipeline crashes.

Again any help most gratefully received and please ask me to clarify if anything is unclear.

Many thanks,

Rob

6

Hi, could you post the crashfile again? Also cc-ing @satra for another pair of eyes

7

Here is the crash file if I put:

for r in range(3):

rather than

for r in range(1):

Thanks again!

File: crash-20170321-095309-k1201869-normalize.a0-69e953ce-e7f0-433e-b813-b5a9635e6871.pklz
Node: metaflow.preproc.normalize.a0
Working directory: /home/k1201869/restingstate2017/workingdir_prepro/metaflow/preproc/_session_id_run1_subject_id_subject001/normalize


Node inputs:

affine_regularization_type = <undefined>
apply_to_files = ['/home/k1201869/restingstate2017/workingdir_prepro/metaflow/preproc/_session_id_run1_subject_id_subject001/coregister/rrun1_reset.nii']
bias_fwhm = <undefined>
bias_regularization = <undefined>
deformation_file = <undefined>
ignore_exception = False
image_to_align = /home/k1201869/restingstate2017/data/subject001/struc/struc_reset.nii
jobtype = estwrite
matlab_cmd = <undefined>
mfile = True
paths = ['/cns_zfs/system/system_ce51_64/spm/spm-12-20141016/']
sampling_distance = <undefined>
smoothness = <undefined>
tpm = <undefined>
use_mcr = <undefined>
use_v8struct = True
warping_regularization = <undefined>
write_bounding_box = <undefined>
write_interp = <undefined>
write_voxel_sizes = [2.0, 2.0, 2.0]



Traceback: 
Traceback (most recent call last):
  File "/home/k1201869/anaconda2/lib/python2.7/site-packages/nipype/pipeline/plugins/base.py", line 543, in _get_result
    raise IOError(error_message)
IOError: Job id (3479096) finished or terminated, but results file does not exist after (5.0) seconds. Batch dir contains crashdump file if node raised an exception.
Node working directory: (/home/k1201869/restingstate2017/workingdir_prepro/metaflow/preproc/_session_id_run1_subject_id_subject001/normalize)
8

@rob this error looks like a problem with the Normalize node. Could you try clearing your working directory and rerunning the script with for r in range(3)?

9

Hi,

Apologies - the crash file I posted was incorrect. The right one is below. The issue seems to be that timings for all 3 runs are being inputted to each run, rather than timings for run1 to run1, timings for run2 to run2 etc.

I imagine the fix may be fairly straightforward but any advice most appreciated

File: crash-20170322-195550-k1201869-modelspec.a0-f96f2687-dfc6-4375-af83-7a313c763f83.pklz
Node: metaflow.l1analysis.modelspec.a0
Working directory: /home/k1201869/restingstate2017/workingdir_prepro/metaflow/l1analysis/_session_id_run1_subject_id_subject001/modelspec


Node inputs:

concatenate_runs = False
event_files = <undefined>
functional_runs = ['/home/k1201869/restingstate2017/workingdir_prepro/metaflow/preproc/_session_id_run1_subject_id_subject001/smooth/swrrun1_reset.nii']
high_pass_filter_cutoff = 128.0
ignore_exception = False
input_units = secs
outlier_files = ['/home/k1201869/restingstate2017/workingdir_prepro/metaflow/preproc/_session_id_run1_subject_id_subject001/art/art.rrun1_reset_outliers.txt']
output_units = secs
realignment_parameters = ['/home/k1201869/restingstate2017/workingdir_prepro/metaflow/preproc/_session_id_run1_subject_id_subject001/realign/rp_run1_reset.txt']
subject_info = [Bunch(amplitudes=None, conditions=['rest', 'control', 'stress'], durations=[[10.5], [88.2], [98.7]], onsets=[[0, 220.5], [10.5, 231], [98.7, 319.2]], pmod=None, regressor_names=None, regressors=None, tmod=None), Bunch(amplitudes=None, conditions=['rest', 'control', 'stress'], durations=[[10.5], [88.2], [98.7]], onsets=[[0, 220.5], [10.5, 231], [98.7, 319.2]], pmod=None, regressor_names=None, regressors=None, tmod=None), Bunch(amplitudes=None, conditions=['rest', 'control', 'stress'], durations=[[10.5], [88.2], [342.3]], onsets=[[0], [10.5], [98.7]], pmod=None, regressor_names=None, regressors=None, tmod=None)]
time_repetition = 2.1



Traceback: 
Traceback (most recent call last):
  File "/home/k1201869/restingstate2017/workingdir_prepro/metaflow/batch/pyscript_20170322_195450_metaflow.l1analysis_modelspec.a0.py", line 32, in <module>
    result = info['node'].run(updatehash=info['updatehash'])
  File "/home/k1201869/anaconda2/envs/MRS_conda/lib/python2.7/site-packages/nipype/pipeline/engine/nodes.py", line 394, in run
    self._run_interface()
  File "/home/k1201869/anaconda2/envs/MRS_conda/lib/python2.7/site-packages/nipype/pipeline/engine/nodes.py", line 504, in _run_interface
    self._result = self._run_command(execute)
  File "/home/k1201869/anaconda2/envs/MRS_conda/lib/python2.7/site-packages/nipype/pipeline/engine/nodes.py", line 630, in _run_command
    result = self._interface.run()
  File "/home/k1201869/anaconda2/envs/MRS_conda/lib/python2.7/site-packages/nipype/interfaces/base.py", line 1043, in run
    runtime = self._run_wrapper(runtime)
  File "/home/k1201869/anaconda2/envs/MRS_conda/lib/python2.7/site-packages/nipype/interfaces/base.py", line 1000, in _run_wrapper
    runtime = self._run_interface(runtime)
  File "/home/k1201869/anaconda2/envs/MRS_conda/lib/python2.7/site-packages/nipype/algorithms/modelgen.py", line 408, in _run_interface
    self._generate_design()
  File "/home/k1201869/anaconda2/envs/MRS_conda/lib/python2.7/site-packages/nipype/algorithms/modelgen.py", line 522, in _generate_design
    super(SpecifySPMModel, self)._generate_design(infolist=infolist)
  File "/home/k1201869/anaconda2/envs/MRS_conda/lib/python2.7/site-packages/nipype/algorithms/modelgen.py", line 402, in _generate_design
    outliers=outliers)
  File "/home/k1201869/anaconda2/envs/MRS_conda/lib/python2.7/site-packages/nipype/algorithms/modelgen.py", line 345, in _generate_standard_design
    sessinfo[i]['scans'] = functional_runs[i]
IndexError: list index out of range
Interface SpecifySPMModel failed to run.
10

Hi,

Was this ever resolved? I am getting this exact same error (‘IndexError: list index out of range’ for sessinfo[i][‘scans’] = functional_runs[i]). I think it has something to do with the fact that only one functional run is listed as an input to the node, but subject info has multiple bunches. Although I could be way off. I’m quite stuck and would appreciate any help!