INTRODUCEREA MOLECULELOR IN FORMAT “SMILES”

library("rcdk")

## Loading required package: fingerprint

library("fingerprint")
smiles <- c('CC(=O)OC1=CC=CC=C1C(O)=O')
mols <- parse.smiles(smiles)
smiles2 <- c('CC(=O)OC1=C(C)C=CC=C1C(O)=O')
mols2 <- parse.smiles(smiles2)
view.molecule.2d(mols[[1]])
view.molecule.2d(mols2 [[1]])

GENERAREA AMPRENTELOR

fp1 <- get.fingerprint(mols[[1]], type='extended')
fp1

## Fingerprint object
##  name =   
##  length =  1024 
##  folded =  FALSE 
##  source =  CDK 
##  bits on =  29 42 54 62 79 89 113 119 123 130 147 151 167 171 174 188 252 267 280 302 417 451 468 478 482 542 545 556 579 606 610 613 645 652 679 713 729 738 742 746 750 760 782 797 811 816 827 841 849 869 927 935 950 953 996 1010 1014

fp2 <- get.fingerprint(mols2[[1]], type='extended')
fp2

## Fingerprint object
##  name =   
##  length =  1024 
##  folded =  FALSE 
##  source =  CDK 
##  bits on =  1 29 42 54 62 79 89 113 119 123 130 147 151 153 167 171 174 188 252 267 280 302 403 417 451 468 478 482 542 545 556 579 606 610 613 645 652 679 713 729 738 742 743 746 750 760 782 797 811 816 827 841 849 869 927 935 950 953 996 1010 1014

fps1 <- lapply(mols, get.fingerprint, type='extended')
fps1

## $`CC(=O)OC1=CC=CC=C1C(O)=O`
## Fingerprint object
##  name =   
##  length =  1024 
##  folded =  FALSE 
##  source =  CDK 
##  bits on =  29 42 54 62 79 89 113 119 123 130 147 151 167 171 174 188 252 267 280 302 417 451 468 478 482 542 545 556 579 606 610 613 645 652 679 713 729 738 742 746 750 760 782 797 811 816 827 841 849 869 927 935 950 953 996 1010 1014

fps2 <- lapply(mols2, get.fingerprint, type='extended')
fps2

## $`CC(=O)OC1=C(C)C=CC=C1C(O)=O`
## Fingerprint object
##  name =   
##  length =  1024 
##  folded =  FALSE 
##  source =  CDK 
##  bits on =  1 29 42 54 62 79 89 113 119 123 130 147 151 153 167 171 174 188 252 267 280 302 403 417 451 468 478 482 542 545 556 579 606 610 613 645 652 679 713 729 738 742 743 746 750 760 782 797 811 816 827 841 849 869 927 935 950 953 996 1010 1014

CALCULUL DISTANTELOR

fp.sim <- fp.sim.matrix(fps1, fps2, method='tanimoto')
fp.sim

##      [,1]
## [1,]    1

fp.dist <- 1 - fp.sim
fp.dist

##      [,1]
## [1,]    0

distance(fp1,fp2, method = 'tanimoto')

## [1] 0.9344262

distance(fp1,fp2, method = 'euclidean')

## [1] 0.998045