Converts mol/sdf into a gspan dataset. Assigns targets from target file to graphs.
Initial idea from https://github.com/amaunz/data-yeast-ac/blob/master/sdf2gsp.pl
- To convert your sdf files and label file into a gspan dataset run:
sdf2gsp.pl <path to sdf files> <sdf file list in correct label order> <targets file>- path to sdf files: Is an absolute path to the folder containing the sdf/mol files
- sdf file list in correct label order: The list of the sdf/mol files in the order that also the labels are ordered
- targets file: the file with the 1-dim targets
Further scripts:
- If you have a multilabel target file, first run
get_dim_x_from_labels.sh <dimension> <multi-dim target file>. This extracts one dimension from every line from a multi-dim target file. split_dataset.pl <training samples>splits a gspan dataset. The training data set has an amount of samples given as argument and the rest of the samples will go into the test set.sdf2gsp_nolabels.pl <path to sdf files>will convert the dataset into gspan without labels (it will simply add a graph id)sdf2gsp_randomlabels.pl <path to sdf files>will convert the dataset into gspan and randomly label each graph with either 1 or -1