Can we train our algorithm on any of these data fields present in the training set? COMPOUND_ID, LIBRARY_ID, BB1_ID, BB2_ID, BB3_ID, TARGET_ID, TARGET_VALUE, NTC_VALUE, ENRICHMENT, LABEL, ECFP4, ECFP6, FCFP4, FCFP6, MACCS, RDK, AVALON, ATOMPAIR, TOPTOR, MW, ALOGP I'd like to clarify if we're restricted to just the fingerprints for algorithm training or not.