MR-induced artifacts on EEG data were removed using a novel algorithm.
First, TruSeq adapters from mate-pair data were removed using Nesoni default parameters (v0.115) (https://github.com/Victorian-Bioinformatics-Consortium/nesoni).
Individuals with more than 30% missing data were removed and loci that were present in less than 80% of samples were discarded.
Approximately seven minutes of data were collected in total, however, the first two minutes of data were removed to account for habituation to the imaging procedure.
Entire locations enclosing gaps and missing data were removed.
Prior to the peer review, personal data were removed, thus ensuring the anonymity of the patients.
To ensure confidentiality, all identifying data were removed, and records were kept under lock and key.
All the zero values of the humming pitch and MIDI data were removed [13 15, 17, 18].
All locations having gaps and missing data were removed from the data set using the complete-deletion option.
Our results showed that the performance decreases gradually as progressively larger amounts of data were removed from the training set.
Rather, we were interested in how the performance progresses as increasing amounts of training data were removed.
