Thanks, Oesteban.
Trying now.
Another thing I want to mention is while using version 1.0.11, if I changed the number of nodes to 16 (mem to 64G), the error actually has gone. And my data set is longitudinal which has two time point (for each subject, we have 1 structural + 2 functional scans for two different days ). Does this mean this error actually relate to memory requirement, and I have to make sure that I have allocate at least 32G memory for each time point of data?
It is very likely that this error is related to memory requirements. If one of the tasks on which that node depends was killed, Python is unaware and nipype could go on without noticing. When this node is submitted to run it finds that no file is present and crashes.