Issue of mutationPositionPlot generation

[liu9756]

Hi , I generated some weird plots of mutationPositionPlot. I use the script of the website, but the generated plot looks like this:
S5_mutationPositionPlot_Tree1.pdf

The code I use is
`mutationPositionPlot<-function(trId){
treeDb<-lng_db_merge%>% filter(!is.na(cloneId),treeId==trId) %>%
mutate(CDR1BeginPosition=0,
FR2BeginPosition=positionInReferenceOfFR2Begin-positionInReferenceOfCDR1Begin,
CDR2BeginPosition=positionInReferenceOfCDR2Begin-positionInReferenceOfCDR1Begin,
FR3BeginPosition=positionInReferenceOfFR3Begin-positionInReferenceOfCDR1Begin,
CDR3BeginPosition=positionInReferenceOfCDR3Begin-positionInReferenceOfCDR1Begin,
FR4BeginPosition=positionInReferenceOfCDR3Begin-positionInReferenceOfCDR1Begin+nchar(nSeqCDR3))

FR2Pos<-treeDb %>%
pull(FR2BeginPosition) %>%
unique()%>% min()

CDR2Pos<-treeDb %>%
pull(CDR2BeginPosition) %>%
unique() %>% min()

FR3Pos<-treeDb %>%
pull(FR3BeginPosition) %>%
unique()%>% min()

CDR3Pos<-treeDb %>%
pull(CDR3BeginPosition) %>%
unique()%>% min()

FR4Pos<-treeDb %>%
pull(FR4BeginPosition) %>%
unique()%>% min()

mutByRegion<-str_extract_all(treeDb %>%
pull(nMutations{CDR1Begin:FR3End}BasedOnGermline),
"[0-9][0-9][0-9]") %>% unlist() %>% table() %>%
enframe(name = "position",value="nMut") %>%
mutate(position=as.integer(position),nMut=as.integer(nMut),
region=case_when(
position<FR2Pos ~ "CDR1",
position<CDR2Pos ~ "FR2",
position<FR3Pos ~ "CDR2",
position<CDR3Pos ~ "FR3"
)) %>%
bind_rows(
str_extract_all(treeDb %>% pull(nMutationsCDR3BasedOnGermline),
"[0-9][0-9][0-9]") %>% unlist() %>% table() %>%
enframe(name = "position",value="nMut") %>%
mutate(position=as.integer(position)+CDR3Pos,nMut=as.integer(nMut),
region="CDR3")
) %>%
bind_rows(
str_extract_all(treeDb %>% pull(nMutationsFR4BasedOnGermline),
"[0-9][0-9][0-9]") %>% unlist() %>% table() %>%
enframe(name = "position",value="nMut") %>%
mutate(position=as.integer(position)+FR4Pos,nMut=as.integer(nMut),
region="FR4")
) %>%
mutate(freqMut=nMut/nrow(treeDb))

g<-mutByRegion%>%
ggplot(aes(x=position,y=freqMut,fill=region)) +
theme_bw()+
geom_bar(stat="identity")+
scale_fill_manual(values=mi_dark)+
geom_vline(xintercept = c(FR2Pos,CDR2Pos,FR3Pos,CDR3Pos,FR4Pos),color="grey28",linetype=2)+
theme(legend.position = "none")+
scale_x_continuous(expand=c(0,0)) +
scale_y_continuous(expand=c(0,0)) +
labs(y="Mutation freq.", x="Position")

print(g)

}

mutationPositionPlot(1)`

My professor said to me this is impossible to have so many mutation of frequency 1.
Could you told me if I correctly use the script or our data didn't fit the script
Thanks!

[liu9756]

I still have another question: I have some plot that looks thicker than others for example tree1 in my sample
S6_mutationPositionPlot_Tree1.pdf
And how do can I make 9 trees into one average tree to compare them to each other? like this

Thanks!

[mizraelson]

HI!

Regarding the first question, it’s hard to say without looking at the actual alignments. It’s not clear how many clones you have in that tree. Technically speaking, it could be just a few clones that all share the same hypermutations or maybe some unidentified alleles.

Again, about the thick plot, it’s difficult to determine what’s happening without reviewing the actual data and knowing how many clones are present.

“And how can I make 9 trees into one average tree to compare them to each other?” Could you clarify what you mean by an average tree?

[liu9756]

Thanks for your reply ! I think I am required to generate a overall plot of shm freq/position for the data, but currently I have several trees, and I am expected to generate a overall plot of them for comparing. Is it possible for the current data I generated from MiXCR to handle this? I have all tsv files that required to generate the mutationPositionPlot.

[mizraelson]

Could you clarify what you mean by the overall plot of trees for comparison? Are you referring to a SHM frequency plot for each tree individually, or do you want to combine the data into a single plot where each tree is represented by a different color? In both cases, you have all the data from MiXCR ready, and it will require some R/Python code to visualize it as you prefer. The script from the website is just an example tailored to a specific case.