R/SimilarityScoreFunctions.R
In Elisa89m/PDSP: Phenotypic Drug Similarity Prediction
#' Structure Similarity compound-CMAPset
#'
#' @description
#' Compute the structure similarity score between the selected compound and the compounds of CMAP
#' @param apset_comp APset object containing the Atom Pair descriptors of new compound(s)
#' @param apsetCMAP APset object containing all Atom Pair descriptors of CMAP compounds
#' @return data.frame (Nx1) with similarity score between new_compound and apsetCMAP
#' @export
AP_simCMAP <-function(apset_comp, apsetCMAP){
# cmp.similarity compute the similarity between atom pairs
AP_comp_CMAP = lapply(seq_along(cid(apsetCMAP)), function(x) cmp.similarity(apset_comp,apsetCMAP[x],mode = 1))# Tanimoto
names(AP_comp_CMAP) = apsetCMAP@ID
new_comp = apset_comp@ID
comp_pairs = OrderCompound(new_comp, apsetCMAP@ID)
StructureDf = data.frame('Structure'= unlist(AP_comp_CMAP), 'row.names' = comp_pairs)
return(StructureDf)
}
#' Structure Similarity between two new compound
#'
#' @description
#' Compute the structure similarity score between the selected compound and the compounds of CMAP
#' @param apset_compA APset object containing the Atom Pair descriptors of new compound
#' @param apset_compB APset object containing the Atom Pair descriptors of new compound
#' @return data.frame containing the similarity score between two selected compounds
#' @export
AP_sim2comp <-function(apset_compA, apset_compB){
pair_score = cmp.similarity(apset_compA,apset_compB,mode = 1)
if(apset_compA@ID < apset_compB@ID){
comp_pair_name = paste0(apset_compA@ID,'_', apset_compB@ID)
} else {
comp_pair_name = paste0(apset_compB@ID, '_', apset_compA@ID)
}
pair_score_df = data.frame('Structure'= pair_score, 'row.names'=comp_pair_name)
return(pair_score_df)
}
#' co-occurrence similarity between two new compound and connectivity map compounds
#'
#' @description
#' Compute the co-occurrence score between the selected compound and the compounds of CMAP
#' @param new_comp_name a character object containing the name associated to the new compound
#' @param PMID_newComps character array containing the PMIDs associated to the new compound
#' @param PMID_cmap list object containing the PMID associated to each compound of CMAP
#' @return data frame containing the compound pair similarity scores between the new compound(s) and all CMAP compounds
#' @export
GetLitteratureCo_occ <-function(new_comp_name,PMID_newComps,PMID_cmap){
commonPMID_newcomp_CMAP = unlist(lapply(names(PMID_cmap), function(comp) length(intersect(PMID_newComps, PMID_cmap[[comp]]))))
names(commonPMID_newcomp_CMAP) = names(PMID_cmap)
unionPMID_newcomp_CMAP = unlist(lapply(names(PMID_cmap), function(comp) length(unique(PMID_newComps, PMID_cmap[[comp]]))))
names(unionPMID_newcomp_CMAP) = names(PMID_cmap)
JI_score_newcomp_CMAP = commonPMID_newcomp_CMAP/unionPMID_newcomp_CMAP
comp_pairs = OrderCompound(new_comp_name,names(PMID_cmap))
PMID_df = data.frame('PMID'= JI_score_newcomp_CMAP, 'row.names' = comp_pairs)
return(PMID_df)
}
#' Gene Expression similarity between two new compound and connecitivity map compounds
#'
#' @description
#' Compute the Correlation score between the selected compound and the compounds of CMAP
#' @param GeneExprNewDrug data.frame containing the gene expression of a new compound
#' @param GeneExprCMAP data.frame containing the gene expression profiles of CMAP
#' @return data.frame object containing the gene expression correlation between the new compound and all CMAP compounds
#' @export
GetGeneExprScore <-function(GeneExprNewDrug,GeneExprCMAP){
new_comp = colnames(GeneExprNewDrug)
common_gene = intersect(rownames(GeneExprNewDrug), rownames(GeneExprCMAP))
CorrelationScore = unlist(lapply(colnames(GeneExprCMAP), function(comp) cor(as.numeric(GeneExprNewDrug[common_gene,new_comp]), as.numeric(GeneExprCMAP[common_gene,comp]))))
names(CorrelationScore) = colnames(GeneExprCMAP)
comp_pairs = OrderCompound(new_comp,colnames(GeneExprCMAP))
CorrelationScoreScaled = (CorrelationScore+ abs(-1))/(abs(-1-1))
GeneExprDF = data.frame('GeneExpr'= CorrelationScoreScaled, 'row.names' = comp_pairs)
return(GeneExprDF)
}
#' ATC code score similarity between a new compound and a set of ATC codes
#'
#' Compute the ATC score between a selected compound and one or more compounds
#' @param newDrugInfo_comp list object containing the information of a new compound
#' @param ATCcode_df data.frame containing the ATC codes for one or more compounds
#' @return data.frame containing the compound pair similarity scores between the ATC code of the new compound and the ATC codes provided in ATCcode_df
#' @export
GetATCcodeScore <-function(newDrugInfo_comp,ATCcode_df){
result_table_freq = data.frame()
new_comp = names(newDrugInfo_comp)
ATCcodeNewDrug = newDrugInfo_comp[[new_comp]][['ATC']]
new_atc_df = data.frame()
i=0
for (atc_code in ATCcodeNewDrug){
i=i+1
row_atc = data.frame('NAME'= new_comp,'L1'= substr(atc_code,1,1), 'L2'= substr(atc_code,1,3), 'L3'=substr(atc_code,1,4), 'L4'=substr(atc_code,1,5), 'L5'=atc_code, 'Number'= i)
new_atc_df= rbind(new_atc_df, row_atc)
}
scores = list('L1'=0.2, 'L2'=0.4, 'L3'=0.6, 'L4'=0.8)
level_type = 1
for(level_atc in c('L1', 'L2', 'L3', 'L4')){
atc_number= 0
level_score_tab = data.frame()
for(atc_codes_level in as.character(new_atc_df[,level_atc] )){
atc_number = atc_number+1
ATCcode_subset = ATCcode_df[which(as.character(ATCcode_df[,level_atc] ) == atc_codes_level),]
if(nrow(ATCcode_subset)>=1){
L_tab_number = lapply(1:nrow(ATCcode_subset), function(row_ind) data.frame('CompA'=paste(ATCcode_subset$NAME[row_ind],ATCcode_subset$Number[row_ind], sep='_'),
'CompB'=paste(new_comp, atc_number,sep='_'), 'LScore'=scores[level_atc], 'L_ext'= atc_codes_level ))
L_tab = do.call("rbind", L_tab_number)
colnames(L_tab)[3:4]=c(level_atc, paste0(level_atc, '_ext'))
level_score_tab = rbind(level_score_tab, L_tab)
}
}
if(level_type==1){
final_table = level_score_tab
} else {
if(nrow(level_score_tab)>=1){
final_table = merge(final_table, level_score_tab, by=c('CompA', 'CompB'), all=TRUE)
}
}
level_type = level_type+1
}
if(nrow(final_table)>=1){
compA = strsplit(as.character(final_table$CompA), '_')
compA = unlist(lapply(compA, function(item) item[[1]][1] ))
compB = strsplit(as.character(final_table$CompB), '_')
compB = unlist(lapply(compB, function(item) item[[1]][1] ))
final_table$CompPair = OrderCompound(compA, compB)
max_value = apply(final_table[,names(scores)], 1, 'max', na.rm=TRUE)
final_table$MAX = max_value
result_table = aggregate(MAX~CompPair, final_table, 'sum')
freq = as.data.frame(table(final_table$CompPair))
colnames(freq)[1] = 'CompPair'
result_table_freq = merge(result_table, freq, by='CompPair')
result_table_freq$mean = result_table_freq$MAX/result_table_freq$Freq
}
return(result_table_freq)
}
#' KEGG data score similarity between a new compound and connecitivity map compounds
#'
#' Compute the KEGG data score between the selected compound and the compounds of CMAP
#' @param newDrugInfo_comp list object containing the information of a new compound
#' @param CMAPData Drug Info CMAP Class object containing the CMAP compounds information
#' @return list object containing three data.frame with the similarity scores based on the Maps, Path and Target Kegg drug information
#' @export
GetKEGGScore <-function(newDrugInfo_comp,CMAPData){
new_comp = names(newDrugInfo_comp)
cmap_maps = CMAPData@Maps
cmap_maps = cmap_maps[which(grepl('^map', cmap_maps))]
cmap_maps = strsplit(cmap_maps, '\\|\\|')
cmap_target = strsplit(CMAPData@Target, '\\|\\|')
cmap_path = strsplit(CMAPData@Path, '\\|\\|')
KEGG_cmap = list('Maps'=cmap_maps, 'Target'= cmap_target, 'Path'= cmap_path)
results = list()
for( Kegg_info in c('Maps','Target', 'Path')){
CMAP_data = KEGG_cmap[[Kegg_info]]
common_info_CMAP_newDrug = unlist(lapply(names(CMAP_data), function(cmap_comp) length(intersect(CMAP_data[[cmap_comp]], newDrugInfo_comp[[new_comp]][[Kegg_info]]))))
union_info_CMAP_newDrug = unlist(lapply(names(CMAP_data), function(cmap_comp) length(unique(c(CMAP_data[[cmap_comp]], newDrugInfo_comp[[new_comp]][[Kegg_info]])))))
JI_kegg_info_CMAP_newDrug = common_info_CMAP_newDrug/union_info_CMAP_newDrug
comp_pairs = OrderCompound(new_comp,names(CMAP_data))
KEGG_info_DF = data.frame( JI_kegg_info_CMAP_newDrug, 'row.names' = comp_pairs)
colnames(KEGG_info_DF) = Kegg_info
results[[paste0('DF_', Kegg_info)]]= na.omit(KEGG_info_DF)
}
return(results)
}
#' All similarity score computation among new compounds
#'
#' Compute all type similarity scores among new compounds
#' @param newDrugInfo list containing the drug information of the new selected compounds
#' @param new_comp_A the name of one drug
#' @param new_comp_B the name of another drug
#' @return list object containing the similairty scores based on all chemical features among two new compounds
#' @export
GetAllSimilarityScores <- function(newDrugInfo, new_comp_A, new_comp_B){
# Structure
structure_score = AP_sim2comp(newDrugInfo[[new_comp_A]][['APsdf']],newDrugInfo[[new_comp_B]][['APsdf']] )
# PMID
JI_PMID = length(intersect(newDrugInfo[[new_comp_A]][['PMID']], newDrugInfo[[new_comp_B]][['PMID']]))/length(unique(c(newDrugInfo[[new_comp_A]][['PMID']],newDrugInfo[[new_comp_B]][['PMID']])))
comp_pair = OrderCompound(new_comp_A,new_comp_B)
JI_PMID_comps = data.frame('PMID'=JI_PMID, row.names=comp_pair)
# GeneExpr
CorrelationScore_df = data.frame()
if ('GeneExpr' %in% names(newDrugInfo[[new_comp_A]]) & 'GeneExpr' %in% names(newDrugInfo[[new_comp_B]])) {
gene_A = rownames(newDrugInfo[[new_comp_A]][['GeneExpr']])
gene_B = rownames(newDrugInfo[[new_comp_B]][['GeneExpr']])
common_genes = intersect(gene_A, gene_B)
CorrelationScore = cor(newDrugInfo[[new_comp_A]][['GeneExpr']][which(gene_A %in% common_genes),1], newDrugInfo[[new_comp_B]][['GeneExpr']][which(gene_A %in% common_genes),1])
CorrelationScore_df = data.frame('GeneExpr'=CorrelationScore, row.names = comp_pair)
}
# KEGG
Kegg_results_new_drug = list()
for( Kegg_info in c('Maps','Target', 'Path')){
KeggData_A = newDrugInfo[[new_comp_A]][[Kegg_info]]
KeggData_B = newDrugInfo[[new_comp_B]][[Kegg_info]]
JI_kegg_info_newDrug = length(intersect(KeggData_A, KeggData_B))/length(unique(c(KeggData_A, KeggData_B)))
KEGG_info_DF = data.frame( JI_kegg_info_newDrug, row.names =comp_pair)
colnames(KEGG_info_DF) = Kegg_info
Kegg_results_new_drug[[paste0('DF_', Kegg_info)]]= KEGG_info_DF
}
# ATC
ATCcodeNewDrug = newDrugInfo[[new_comp_B]][['ATC']]
new_atc_df_B = data.frame()
i=0
for (atc_code in ATCcodeNewDrug){
i=i+1
row_atc = data.frame('NAME'= new_comp_B,'L1'= substr(atc_code,1,1), 'L2'= substr(atc_code,1,3), 'L3'=substr(atc_code,1,4), 'L4'=substr(atc_code,1,5), 'L5'=atc_code, 'Number'= i)
new_atc_df_B= rbind(new_atc_df_B, row_atc)
}
Score_twoComps = GetATCcodeScore(newDrugInfo[new_comp_A], new_atc_df_B)
return(list('Structure'=structure_score, 'GeneExpr'=CorrelationScore_df, 'PMID'= JI_PMID_comps, 'Maps'=Kegg_results_new_drug$DF_Maps,
'Target'=Kegg_results_new_drug$DF_Target, 'Path'=Kegg_results_new_drug$DF_Path,'ATC'= Score_twoComps))
}
#' All similarity score computation between two new compound and connecitivity map compounds
#'
#' Compute the KEGG data score between the selected compound and the compounds of CMAP
#' @param CMAPData Drug Info CMAP Class object containing the CMAP compounds information
#' @param newDrugInfo list containing the drug information of the new selected compounds
#' @return list object containing the similairty scores based on all chemical features between new compounds and CMAP compounds
#' @export
GetAllSimilarityScoresCMAP <- function(CMAPData, newDrugInfo){
StructureDF = data.frame()
PMID_DF = data.frame()
GeneExprDF = data.frame()
Path_DF = data.frame()
Maps_DF = data.frame()
Target_DF = data.frame()
ATC_DF = data.frame()
apsetCMAP = CMAPData@APsdf[[1]]
PMID_cmap = CMAPData@pmid
for (new_comp in names(newDrugInfo)){
print(new_comp)
# Structure similarity Score computation
apset_new_comp = newDrugInfo[[new_comp]][['APsdf']]
StructureDFComp = AP_simCMAP(apset_new_comp,apsetCMAP)
StructureDF = rbind(StructureDF,StructureDFComp)
# Co-occurrence similarity Score computation
PMID_newComps = newDrugInfo[[new_comp]]['PMID']
PMID_DFComp = GetLitteratureCo_occ(new_comp,PMID_newComps,PMID_cmap)
PMID_DF = rbind(PMID_DF,PMID_DFComp)
# Gene Expression Similarity Score computation
if('GeneExpr' %in% names(newDrugInfo[[new_comp]])){
GeneExprCMAP = CMAPData@GeneExpr
GeneExprNewDrug = newDrugInfo[[new_comp]][['GeneExpr']]
GeneExprDFComp = GetGeneExprScore(GeneExprNewDrug, GeneExprCMAP)
GeneExprDF = rbind(GeneExprDF, GeneExprDFComp)
}
# KEGG data similarity Score computation
KEGG_DF_list = GetKEGGScore(newDrugInfo[new_comp], CMAPData)
Path_DF = rbind(Path_DF,KEGG_DF_list$DF_Path)
Maps_DF = rbind(Maps_DF,KEGG_DF_list$DF_Maps)
Target_DF = rbind(Target_DF,KEGG_DF_list$DF_Target)
# ATC similarity Score computation
ATCcodeCMAP = CMAPData@ATC
ATC_score_DF = GetATCcodeScore(newDrugInfo[new_comp],ATCcodeCMAP)
ATC_DF = rbind(ATC_DF, ATC_score_DF)
}
################
# Score Between new compounds
new_comps= names(newDrugInfo)
number_comps = length(new_comps)
i=0
for (new_comp_A in new_comps){
i=i+1
if(i!=number_comps){
for (new_comp_B in new_comps[(i+1):number_comps]){
list_scores_AB = GetAllSimilarityScores(newDrugInfo, new_comp_A ,new_comp_B)
StructureDF = rbind(StructureDF,list_scores_AB$Structure )
PMID_DF = rbind(PMID_DF, list_scores_AB$PMID)
GeneExprDF = rbind(GeneExprDF, list_scores_AB$GeneExpr)
Path_DF = rbind(Path_DF, list_scores_AB$Path)
Maps_DF = rbind(Maps_DF, list_scores_AB$Maps)
Target_DF = rbind(Target_DF, list_scores_AB$Target)
ATC_DF = rbind(ATC_DF, list_scores_AB$ATC )
}}
}
colnames(ATC_DF)[4] = 'ATC'
rownames(ATC_DF) = ATC_DF$CompPair
return(list('Structure'=StructureDF, 'GeneExpr'=GeneExprDF, 'PMID'= PMID_DF, 'Maps'=Maps_DF, 'Target'=Target_DF, 'Path'=Path_DF,'ATC'=ATC_DF))
}
Elisa89m/PDSP documentation built on May 30, 2019, 1:26 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
#' Structure Similarity compound-CMAPset
#'
#' @description
#' Compute the structure similarity score between the selected compound and the compounds of CMAP
#' @param apset_comp APset object containing the Atom Pair descriptors of new compound(s)
#' @param apsetCMAP APset object containing all Atom Pair descriptors of CMAP compounds
#' @return data.frame (Nx1) with similarity score between new_compound and apsetCMAP
#' @export
AP_simCMAP <-function(apset_comp, apsetCMAP){
# cmp.similarity compute the similarity between atom pairs
AP_comp_CMAP = lapply(seq_along(cid(apsetCMAP)), function(x) cmp.similarity(apset_comp,apsetCMAP[x],mode = 1))# Tanimoto
names(AP_comp_CMAP) = apsetCMAP@ID
new_comp = apset_comp@ID
comp_pairs = OrderCompound(new_comp, apsetCMAP@ID)
StructureDf = data.frame('Structure'= unlist(AP_comp_CMAP), 'row.names' = comp_pairs)
return(StructureDf)
}
#' Structure Similarity between two new compound
#'
#' @description
#' Compute the structure similarity score between the selected compound and the compounds of CMAP
#' @param apset_compA APset object containing the Atom Pair descriptors of new compound
#' @param apset_compB APset object containing the Atom Pair descriptors of new compound
#' @return data.frame containing the similarity score between two selected compounds
#' @export
AP_sim2comp <-function(apset_compA, apset_compB){
pair_score = cmp.similarity(apset_compA,apset_compB,mode = 1)
if(apset_compA@ID < apset_compB@ID){
comp_pair_name = paste0(apset_compA@ID,'_', apset_compB@ID)
} else {
comp_pair_name = paste0(apset_compB@ID, '_', apset_compA@ID)
}
pair_score_df = data.frame('Structure'= pair_score, 'row.names'=comp_pair_name)
return(pair_score_df)
}
#' co-occurrence similarity between two new compound and connectivity map compounds
#'
#' @description
#' Compute the co-occurrence score between the selected compound and the compounds of CMAP
#' @param new_comp_name a character object containing the name associated to the new compound
#' @param PMID_newComps character array containing the PMIDs associated to the new compound
#' @param PMID_cmap list object containing the PMID associated to each compound of CMAP
#' @return data frame containing the compound pair similarity scores between the new compound(s) and all CMAP compounds
#' @export
GetLitteratureCo_occ <-function(new_comp_name,PMID_newComps,PMID_cmap){
commonPMID_newcomp_CMAP = unlist(lapply(names(PMID_cmap), function(comp) length(intersect(PMID_newComps, PMID_cmap[[comp]]))))
names(commonPMID_newcomp_CMAP) = names(PMID_cmap)
unionPMID_newcomp_CMAP = unlist(lapply(names(PMID_cmap), function(comp) length(unique(PMID_newComps, PMID_cmap[[comp]]))))
names(unionPMID_newcomp_CMAP) = names(PMID_cmap)
JI_score_newcomp_CMAP = commonPMID_newcomp_CMAP/unionPMID_newcomp_CMAP
comp_pairs = OrderCompound(new_comp_name,names(PMID_cmap))
PMID_df = data.frame('PMID'= JI_score_newcomp_CMAP, 'row.names' = comp_pairs)
return(PMID_df)
}
#' Gene Expression similarity between two new compound and connecitivity map compounds
#'
#' @description
#' Compute the Correlation score between the selected compound and the compounds of CMAP
#' @param GeneExprNewDrug data.frame containing the gene expression of a new compound
#' @param GeneExprCMAP data.frame containing the gene expression profiles of CMAP
#' @return data.frame object containing the gene expression correlation between the new compound and all CMAP compounds
#' @export
GetGeneExprScore <-function(GeneExprNewDrug,GeneExprCMAP){
new_comp = colnames(GeneExprNewDrug)
common_gene = intersect(rownames(GeneExprNewDrug), rownames(GeneExprCMAP))
CorrelationScore = unlist(lapply(colnames(GeneExprCMAP), function(comp) cor(as.numeric(GeneExprNewDrug[common_gene,new_comp]), as.numeric(GeneExprCMAP[common_gene,comp]))))
names(CorrelationScore) = colnames(GeneExprCMAP)
comp_pairs = OrderCompound(new_comp,colnames(GeneExprCMAP))
CorrelationScoreScaled = (CorrelationScore+ abs(-1))/(abs(-1-1))
GeneExprDF = data.frame('GeneExpr'= CorrelationScoreScaled, 'row.names' = comp_pairs)
return(GeneExprDF)
}
#' ATC code score similarity between a new compound and a set of ATC codes
#'
#' Compute the ATC score between a selected compound and one or more compounds
#' @param newDrugInfo_comp list object containing the information of a new compound
#' @param ATCcode_df data.frame containing the ATC codes for one or more compounds
#' @return data.frame containing the compound pair similarity scores between the ATC code of the new compound and the ATC codes provided in ATCcode_df
#' @export
GetATCcodeScore <-function(newDrugInfo_comp,ATCcode_df){
result_table_freq = data.frame()
new_comp = names(newDrugInfo_comp)
ATCcodeNewDrug = newDrugInfo_comp[[new_comp]][['ATC']]
new_atc_df = data.frame()
i=0
for (atc_code in ATCcodeNewDrug){
i=i+1
row_atc = data.frame('NAME'= new_comp,'L1'= substr(atc_code,1,1), 'L2'= substr(atc_code,1,3), 'L3'=substr(atc_code,1,4), 'L4'=substr(atc_code,1,5), 'L5'=atc_code, 'Number'= i)
new_atc_df= rbind(new_atc_df, row_atc)
}
scores = list('L1'=0.2, 'L2'=0.4, 'L3'=0.6, 'L4'=0.8)
level_type = 1
for(level_atc in c('L1', 'L2', 'L3', 'L4')){
atc_number= 0
level_score_tab = data.frame()
for(atc_codes_level in as.character(new_atc_df[,level_atc] )){
atc_number = atc_number+1
ATCcode_subset = ATCcode_df[which(as.character(ATCcode_df[,level_atc] ) == atc_codes_level),]
if(nrow(ATCcode_subset)>=1){
L_tab_number = lapply(1:nrow(ATCcode_subset), function(row_ind) data.frame('CompA'=paste(ATCcode_subset$NAME[row_ind],ATCcode_subset$Number[row_ind], sep='_'),
'CompB'=paste(new_comp, atc_number,sep='_'), 'LScore'=scores[level_atc], 'L_ext'= atc_codes_level ))
L_tab = do.call("rbind", L_tab_number)
colnames(L_tab)[3:4]=c(level_atc, paste0(level_atc, '_ext'))
level_score_tab = rbind(level_score_tab, L_tab)
}
}
if(level_type==1){
final_table = level_score_tab
} else {
if(nrow(level_score_tab)>=1){
final_table = merge(final_table, level_score_tab, by=c('CompA', 'CompB'), all=TRUE)
}
}
level_type = level_type+1
}
if(nrow(final_table)>=1){
compA = strsplit(as.character(final_table$CompA), '_')
compA = unlist(lapply(compA, function(item) item[[1]][1] ))
compB = strsplit(as.character(final_table$CompB), '_')
compB = unlist(lapply(compB, function(item) item[[1]][1] ))
final_table$CompPair = OrderCompound(compA, compB)
max_value = apply(final_table[,names(scores)], 1, 'max', na.rm=TRUE)
final_table$MAX = max_value
result_table = aggregate(MAX~CompPair, final_table, 'sum')
freq = as.data.frame(table(final_table$CompPair))
colnames(freq)[1] = 'CompPair'
result_table_freq = merge(result_table, freq, by='CompPair')
result_table_freq$mean = result_table_freq$MAX/result_table_freq$Freq
}
return(result_table_freq)
}
#' KEGG data score similarity between a new compound and connecitivity map compounds
#'
#' Compute the KEGG data score between the selected compound and the compounds of CMAP
#' @param newDrugInfo_comp list object containing the information of a new compound
#' @param CMAPData Drug Info CMAP Class object containing the CMAP compounds information
#' @return list object containing three data.frame with the similarity scores based on the Maps, Path and Target Kegg drug information
#' @export
GetKEGGScore <-function(newDrugInfo_comp,CMAPData){
new_comp = names(newDrugInfo_comp)
cmap_maps = CMAPData@Maps
cmap_maps = cmap_maps[which(grepl('^map', cmap_maps))]
cmap_maps = strsplit(cmap_maps, '\\|\\|')
cmap_target = strsplit(CMAPData@Target, '\\|\\|')
cmap_path = strsplit(CMAPData@Path, '\\|\\|')
KEGG_cmap = list('Maps'=cmap_maps, 'Target'= cmap_target, 'Path'= cmap_path)
results = list()
for( Kegg_info in c('Maps','Target', 'Path')){
CMAP_data = KEGG_cmap[[Kegg_info]]
common_info_CMAP_newDrug = unlist(lapply(names(CMAP_data), function(cmap_comp) length(intersect(CMAP_data[[cmap_comp]], newDrugInfo_comp[[new_comp]][[Kegg_info]]))))
union_info_CMAP_newDrug = unlist(lapply(names(CMAP_data), function(cmap_comp) length(unique(c(CMAP_data[[cmap_comp]], newDrugInfo_comp[[new_comp]][[Kegg_info]])))))
JI_kegg_info_CMAP_newDrug = common_info_CMAP_newDrug/union_info_CMAP_newDrug
comp_pairs = OrderCompound(new_comp,names(CMAP_data))
KEGG_info_DF = data.frame( JI_kegg_info_CMAP_newDrug, 'row.names' = comp_pairs)
colnames(KEGG_info_DF) = Kegg_info
results[[paste0('DF_', Kegg_info)]]= na.omit(KEGG_info_DF)
}
return(results)
}
#' All similarity score computation among new compounds
#'
#' Compute all type similarity scores among new compounds
#' @param newDrugInfo list containing the drug information of the new selected compounds
#' @param new_comp_A the name of one drug
#' @param new_comp_B the name of another drug
#' @return list object containing the similairty scores based on all chemical features among two new compounds
#' @export
GetAllSimilarityScores <- function(newDrugInfo, new_comp_A, new_comp_B){
# Structure
structure_score = AP_sim2comp(newDrugInfo[[new_comp_A]][['APsdf']],newDrugInfo[[new_comp_B]][['APsdf']] )
# PMID
JI_PMID = length(intersect(newDrugInfo[[new_comp_A]][['PMID']], newDrugInfo[[new_comp_B]][['PMID']]))/length(unique(c(newDrugInfo[[new_comp_A]][['PMID']],newDrugInfo[[new_comp_B]][['PMID']])))
comp_pair = OrderCompound(new_comp_A,new_comp_B)
JI_PMID_comps = data.frame('PMID'=JI_PMID, row.names=comp_pair)
# GeneExpr
CorrelationScore_df = data.frame()
if ('GeneExpr' %in% names(newDrugInfo[[new_comp_A]]) & 'GeneExpr' %in% names(newDrugInfo[[new_comp_B]])) {
gene_A = rownames(newDrugInfo[[new_comp_A]][['GeneExpr']])
gene_B = rownames(newDrugInfo[[new_comp_B]][['GeneExpr']])
common_genes = intersect(gene_A, gene_B)
CorrelationScore = cor(newDrugInfo[[new_comp_A]][['GeneExpr']][which(gene_A %in% common_genes),1], newDrugInfo[[new_comp_B]][['GeneExpr']][which(gene_A %in% common_genes),1])
CorrelationScore_df = data.frame('GeneExpr'=CorrelationScore, row.names = comp_pair)
}
# KEGG
Kegg_results_new_drug = list()
for( Kegg_info in c('Maps','Target', 'Path')){
KeggData_A = newDrugInfo[[new_comp_A]][[Kegg_info]]
KeggData_B = newDrugInfo[[new_comp_B]][[Kegg_info]]
JI_kegg_info_newDrug = length(intersect(KeggData_A, KeggData_B))/length(unique(c(KeggData_A, KeggData_B)))
KEGG_info_DF = data.frame( JI_kegg_info_newDrug, row.names =comp_pair)
colnames(KEGG_info_DF) = Kegg_info
Kegg_results_new_drug[[paste0('DF_', Kegg_info)]]= KEGG_info_DF
}
# ATC
ATCcodeNewDrug = newDrugInfo[[new_comp_B]][['ATC']]
new_atc_df_B = data.frame()
i=0
for (atc_code in ATCcodeNewDrug){
i=i+1
row_atc = data.frame('NAME'= new_comp_B,'L1'= substr(atc_code,1,1), 'L2'= substr(atc_code,1,3), 'L3'=substr(atc_code,1,4), 'L4'=substr(atc_code,1,5), 'L5'=atc_code, 'Number'= i)
new_atc_df_B= rbind(new_atc_df_B, row_atc)
}
Score_twoComps = GetATCcodeScore(newDrugInfo[new_comp_A], new_atc_df_B)
return(list('Structure'=structure_score, 'GeneExpr'=CorrelationScore_df, 'PMID'= JI_PMID_comps, 'Maps'=Kegg_results_new_drug$DF_Maps,
'Target'=Kegg_results_new_drug$DF_Target, 'Path'=Kegg_results_new_drug$DF_Path,'ATC'= Score_twoComps))
}
#' All similarity score computation between two new compound and connecitivity map compounds
#'
#' Compute the KEGG data score between the selected compound and the compounds of CMAP
#' @param CMAPData Drug Info CMAP Class object containing the CMAP compounds information
#' @param newDrugInfo list containing the drug information of the new selected compounds
#' @return list object containing the similairty scores based on all chemical features between new compounds and CMAP compounds
#' @export
GetAllSimilarityScoresCMAP <- function(CMAPData, newDrugInfo){
StructureDF = data.frame()
PMID_DF = data.frame()
GeneExprDF = data.frame()
Path_DF = data.frame()
Maps_DF = data.frame()
Target_DF = data.frame()
ATC_DF = data.frame()
apsetCMAP = CMAPData@APsdf[[1]]
PMID_cmap = CMAPData@pmid
for (new_comp in names(newDrugInfo)){
print(new_comp)
# Structure similarity Score computation
apset_new_comp = newDrugInfo[[new_comp]][['APsdf']]
StructureDFComp = AP_simCMAP(apset_new_comp,apsetCMAP)
StructureDF = rbind(StructureDF,StructureDFComp)
# Co-occurrence similarity Score computation
PMID_newComps = newDrugInfo[[new_comp]]['PMID']
PMID_DFComp = GetLitteratureCo_occ(new_comp,PMID_newComps,PMID_cmap)
PMID_DF = rbind(PMID_DF,PMID_DFComp)
# Gene Expression Similarity Score computation
if('GeneExpr' %in% names(newDrugInfo[[new_comp]])){
GeneExprCMAP = CMAPData@GeneExpr
GeneExprNewDrug = newDrugInfo[[new_comp]][['GeneExpr']]
GeneExprDFComp = GetGeneExprScore(GeneExprNewDrug, GeneExprCMAP)
GeneExprDF = rbind(GeneExprDF, GeneExprDFComp)
}
# KEGG data similarity Score computation
KEGG_DF_list = GetKEGGScore(newDrugInfo[new_comp], CMAPData)
Path_DF = rbind(Path_DF,KEGG_DF_list$DF_Path)
Maps_DF = rbind(Maps_DF,KEGG_DF_list$DF_Maps)
Target_DF = rbind(Target_DF,KEGG_DF_list$DF_Target)
# ATC similarity Score computation
ATCcodeCMAP = CMAPData@ATC
ATC_score_DF = GetATCcodeScore(newDrugInfo[new_comp],ATCcodeCMAP)
ATC_DF = rbind(ATC_DF, ATC_score_DF)
}
################
# Score Between new compounds
new_comps= names(newDrugInfo)
number_comps = length(new_comps)
i=0
for (new_comp_A in new_comps){
i=i+1
if(i!=number_comps){
for (new_comp_B in new_comps[(i+1):number_comps]){
list_scores_AB = GetAllSimilarityScores(newDrugInfo, new_comp_A ,new_comp_B)
StructureDF = rbind(StructureDF,list_scores_AB$Structure )
PMID_DF = rbind(PMID_DF, list_scores_AB$PMID)
GeneExprDF = rbind(GeneExprDF, list_scores_AB$GeneExpr)
Path_DF = rbind(Path_DF, list_scores_AB$Path)
Maps_DF = rbind(Maps_DF, list_scores_AB$Maps)
Target_DF = rbind(Target_DF, list_scores_AB$Target)
ATC_DF = rbind(ATC_DF, list_scores_AB$ATC )
}}
}
colnames(ATC_DF)[4] = 'ATC'
rownames(ATC_DF) = ATC_DF$CompPair
return(list('Structure'=StructureDF, 'GeneExpr'=GeneExprDF, 'PMID'= PMID_DF, 'Maps'=Maps_DF, 'Target'=Target_DF, 'Path'=Path_DF,'ATC'=ATC_DF))
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.