Mercurial > repos > davidvanzessen > clonal_sequences_in_paired_samples

comparison RScript.r @ 36:d592dab2fca1 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author davidvanzessen
date Mon, 31 Aug 2015 09:28:38 -0400
parents 32d8a5abed4c
children 623bbe972363
comparison
equal deleted inserted replaced
35:32d8a5abed4c 36:d592dab2fca1
154 merge.list = patientMerge$merge 154 merge.list = patientMerge$merge
155 155
156 patient1.fuzzy = patient1[!(patient1$merge %in% merge.list),] 156 patient1.fuzzy = patient1[!(patient1$merge %in% merge.list),]
157 patient2.fuzzy = patient2[!(patient2$merge %in% merge.list),] 157 patient2.fuzzy = patient2[!(patient2$merge %in% merge.list),]
158 158
159 patient1.fuzzy$merge = paste(patient1.fuzzy$V_Segment_Major_Gene, patient1.fuzzy$J_Segment_Major_Gene, patient1.fuzzy$CDR3_Sense_Sequence) 159 #patient1.fuzzy$merge = paste(patient1.fuzzy$V_Segment_Major_Gene, patient1.fuzzy$J_Segment_Major_Gene, patient1.fuzzy$CDR3_Sense_Sequence)
160 patient2.fuzzy$merge = paste(patient2.fuzzy$V_Segment_Major_Gene, patient2.fuzzy$J_Segment_Major_Gene, patient2.fuzzy$CDR3_Sense_Sequence) 160 #patient2.fuzzy$merge = paste(patient2.fuzzy$V_Segment_Major_Gene, patient2.fuzzy$J_Segment_Major_Gene, patient2.fuzzy$CDR3_Sense_Sequence)
161
162 patient1.fuzzy$merge = paste(patient1.fuzzy$locus_V, patient1.fuzzy$locus_J, patient1.fuzzy$CDR3_Sense_Sequence)
163 patient2.fuzzy$merge = paste(patient2.fuzzy$locus_V, patient2.fuzzy$locus_J, patient2.fuzzy$CDR3_Sense_Sequence)
161 164
162 merge.freq.table = data.frame(table(c(patient1.fuzzy$merge, patient2.fuzzy$merge))) 165 merge.freq.table = data.frame(table(c(patient1.fuzzy$merge, patient2.fuzzy$merge)))
163 merge.freq.table.gt.1 = merge.freq.table[merge.freq.table$Freq > 1,] 166 merge.freq.table.gt.1 = merge.freq.table[merge.freq.table$Freq > 1,]
164 167
165 patient1.fuzzy = patient1.fuzzy[patient1.fuzzy$merge %in% merge.freq.table.gt.1$Var1,] 168 patient1.fuzzy = patient1.fuzzy[patient1.fuzzy$merge %in% merge.freq.table.gt.1$Var1,]
169 current.merge.1 = patient1.fuzzy[1,"merge"] 172 current.merge.1 = patient1.fuzzy[1,"merge"]
170 current.clone.seq.1 = patient1.fuzzy[1,"Clone_Sequence"] 173 current.clone.seq.1 = patient1.fuzzy[1,"Clone_Sequence"]
171 current.merge.in.2 = patient2.fuzzy[patient2.fuzzy$merge == current.merge.1,] 174 current.merge.in.2 = patient2.fuzzy[patient2.fuzzy$merge == current.merge.1,]
172 175
173 #agrep/adist the two samples 176 #agrep/adist the two samples
174 agrep.match = agrep(current.clone.seq.1, current.merge.in.2$Clone_Sequence, max.distance = 1, costs=list(insertions=0.1, deletions=0.1, substitutions=1)) 177 agrep.match = agrep(current.clone.seq.1, current.merge.in.2$Clone_Sequence, max.distance = 9, costs=list(insertions=1, deletions=1, substitutions=10))
175 178
176 179
177 if(length(agrep.match) == 1){ 180 if(length(agrep.match) == 1){
178 current.clone.seq.2 = patient2.fuzzy[agrep.match,"Clone_Sequence"] 181 current.clone.seq.2 = patient2.fuzzy[agrep.match,"Clone_Sequence"]
179 patientMerge.new.row = data.frame(merge=current.clone.seq.1, 182 patientMerge.new.row = data.frame(merge=current.clone.seq.1,
231 patient2.fuzzy <<- patient2.fuzzy[-c(agrep.match),] 234 patient2.fuzzy <<- patient2.fuzzy[-c(agrep.match),]
232 235
233 236
234 } else if (length(agrep.match) > 1){ 237 } else if (length(agrep.match) > 1){
235 #multiple matches, whatdo? 238 #multiple matches, whatdo?
236 cat(paste("<tr><td>", "Multiple matches found for ", current.merge.1, " in ", patient, "</td></tr>", sep=""), file=logfile, append=T) 239 cat(paste("<tr><td>", "Multiple matches found for ", current.merge.1, ", ", current.clone.seq.1, " in ", patient, ", ", oneSample, "</td></tr>", sep=""), file=logfile, append=T)
237 } 240 }
238 patient1.fuzzy = patient1.fuzzy[-1,] 241 patient1.fuzzy = patient1.fuzzy[-1,]
239 } 242 }
240 243
241 #adist(patient1.fuzzy$Clone_Sequence, patient2.fuzzy$Clone_Sequence, list(insertions=0.1, deletions=0.1, substitutions=1)) 244 #adist(patient1.fuzzy$Clone_Sequence, patient2.fuzzy$Clone_Sequence, list(insertions=0.1, deletions=0.1, substitutions=1))