Hi
I am using MOE-Knime extension for QSAR study. After calculating fingerprints from moe, I dont know, how can i separate them into individual columns for each fingerprints. otherswise all the fingerprints are in the same column. Can any one help in this ?
Nathan
Hi Nathan,
I'm not sure what you mean by "are in the same column". If you calculate different fingerprints they should come up in different columns. These columns are of the deprecated type BitvectorCell in version 1.2.0. This will change in the near future!
In Version 2.0.0 the fingerprint node will export SparseBitVectorCells. In order to separate the individual bits you would need nodes that work on these cell types. In general I would not recommend processing MOE fingerprints as mere bitvectors as for some of them the encoding as bitvectors is not free of information-loss. To calculate similarities I would highly recommend using the fingerprint model node!
For QSAR studies you might like to use the MACCS keys as descriptors. In version 2.0.0 there will be a special node for calculating these keys (and also a substructure counter node). In the menatime you can calculate the MACCS key if you copy the file q_maccs.svl from $MOE/sample into $MOE/lib/svl/custom. Then use the QSAR node and select "Other 2D descriptors".
Regards
Guido
Hi Nathan,
Use rename node to convert bitvector fingerprint column to string format, to get 1's and 0's, then use 2 string replacer nodes to replace 0 by 0, and 1 by 1, . Next, use the cell splitter node to split by . The values will be integers.
This sequence of nodes should work with any fingerprint.
split by comma ,