Neuron_search
Neuron Search fig
Panels for Neuron Search figure.
neuron_search_score_mat <- load_si_data("neuron_search_score_mat.rds")
options(flycircuit.scoremat="neuron_search_score_mat")Scale bar
imscalebar(seq(0,1,len=11),col=jet.colors(10),horizontal=F, mar=c(0,0,0,0))
Panel B: Pair of same raw image, different segmentation
Plot the query neuron
nopen3d()## glX
## 4plot3d(FCWB)
op=structure(list(FOV = 0, userMatrix = structure(c(1, 0, 0, 0, 0, -1, 0, 0, 0, 0, -1, 0, 0, 0, 0, 1), .Dim = c(4L, 4L)), scale = c(1, 1, 1), zoom = 0.218791082501411), .Names = c("FOV", "userMatrix", "scale", "zoom"))
par3d(op)
plot3d("FruMARCM-M001629_seg001", col='black', lwd=4, soma=T)
Plot the query and the top hit.
clear3d()
par3d(op)
par3d(zoom=0.631736993789673)
plot3d("FruMARCM-M001629_seg001", col='black', lwd=1, soma=T)
plot3d(names(sort(fc_nblast("FruMARCM-M001629_seg001"), decreasing=T)[2]), col='red', lwd=4, soma=T)
Plot the top 8.
clear3d()
par3d(op)
par3d(zoom=0.631736993789673)
plot3d("FruMARCM-M001629_seg001", col='black', lwd=1, soma=T)
plot3d(names(sort(fc_nblast("FruMARCM-M001629_seg001"), decreasing=T)[2]), col='red', lwd=4, soma=T)
plot3d(names(sort(fc_nblast("FruMARCM-M001629_seg001"), decreasing=T)[3:11]), soma=T)
For C panels
Plot histogram of scores, showing all hits. It is an inset in another graph.
par(mar=c(0,2,0,0))
qq=qplot(fc_nblast("FruMARCM-M001629_seg001"), binwidth=200, xlab="Forward score", ylab="Count")
qq+ theme(panel.background=element_rect(fill="grey80"), panel.grid.minor=element_line(colour='grey80'), axis.text.x=element_text(size=22), axis.text.y=element_text(size=22), axis.title.x=element_text(size=26), axis.title.y=element_text(size=26)) + geom_vline(xintercept = 0, colour="red", linetype=2, size=1)
# the probability for values under 0 is:
fwdfru=sort(fc_nblast("FruMARCM-M001629_seg001"), decreasing=T)
ecdf(fwdfru)(0)## [1] 0.9652179Plot histogram with xlim at -5000 in zoomed version.
same2zoom=qplot(fc_nblast("FruMARCM-M001629_seg001"), bindwidth=60, ylim=c(0,10), xlim=c(6500, 10000), xlab="Forward score", ylab="Count") + theme(panel.background=element_rect(fill="grey80"), panel.grid.minor=element_line(colour='grey80'), axis.text.x=element_text(size=22), axis.text.y=element_text(size=22), axis.title.x=element_text(size=26), axis.title.y=element_text(size=26))
par(mar=c(0,2,0,0))
plot(same2zoom,lwd=.3)## Warning: position_stack requires constant width: output may be incorrect
Show example search with auditory neuron with raw scores.
Histogram of scores coloured by score bins.
fwdfru_5000=fwdfru[fwdfru>-5000]
bb2=cut(fwdfru_5000, breaks=c(-5000,seq(0, max(fwdfru), len=10)))
cols2=mixOmics::jet.colors(10)[bb2]
names(cols2)=names(fwdfru_5000)
m<-cbind(fwdfru_5000,cols2)
m<-as.data.frame(m, stringsAsFactors=F)
m$cols2=factor(m$cols2)
rownames(m)=names(fwdfru_5000)
#Replace "#00008E" (dark blue) for "#000000" (black)
levels(m$cols2)=c("#000000", "#0000FF", "#0070FF", "#00DFFF", "#50FFAE", "#7F0000", "#BFFF3F", "#FF5F00", "#FFCE00")
qq=qplot(fwdfru_5000, geom="histogram", binwidth=180, xlab="Foward score", ylab="Count", colour=factor(cols2), fill=factor(cols2))
scoreplot=qq+scale_colour_manual(values=c("#000000", "#0000FF", "#0070FF", "#00DFFF", "#50FFAE", "#7F0000", "#BFFF3F", "#FF5F00", "#FFCE00"))+scale_fill_manual(values=c("#000000", "#0000FF", "#0070FF", "#00DFFF", "#50FFAE", "#7F0000", "#BFFF3F", "#FF5F00", "#FFCE00"))+ theme(axis.text.x=element_text(size=22), axis.text.y=element_text(size=22), axis.title.x=element_text(size=26), axis.title.y=element_text(size=26), legend.position="none")
par(mar=c(0,2,0,0))
plot(scoreplot)
Make neuron plots of each coloured group in the histogram (score bins), except for last group (in black, <0).
First group (red)
clear3d()
op=structure(list(FOV = 0, userMatrix = structure(c(1, 0, 0, 0, 0, -1, 0, 0, 0, 0, -1, 0, 0, 0, 0, 1), .Dim = c(4L, 4L)), scale = c(1, 1, 1), zoom = 0.633195042610168), .Names = c("FOV", "userMatrix", "scale", "zoom"))
par3d(op)
plot3d(rownames(subset(m, cols2==as.character(unique(m$cols2)[1]))), soma=T, col=as.character(unique(m$cols2)[1]), lwd=2)
Second group.
clear3d()
par3d(op)
plot3d(rownames(subset(m, cols2==as.character(unique(m$cols2)[2]))), soma=T, col=as.character(unique(m$cols2)[2]))
Third group.
clear3d()
par3d(op)
plot3d(rownames(subset(m, cols2==as.character(unique(m$cols2)[3]))), soma=T, col=as.character(unique(m$cols2)[3]))
Fourth group.
clear3d()
par3d(op, zoom=0.6648548)
plot3d(rownames(subset(m, cols2==as.character(unique(m$cols2)[4]))), soma=T, col=as.character(unique(m$cols2)[4]))
Fifth group.
clear3d()
par3d(op, zoom=0.6648548)
plot3d(rownames(subset(m, cols2==as.character(unique(m$cols2)[5]))), soma=T, col=as.character(unique(m$cols2)[5]))
Sixth group.
clear3d()
par3d(op)
plot3d(rownames(subset(m, cols2==as.character(unique(m$cols2)[6]))), soma=T, col=as.character(unique(m$cols2)[6]))
Seventh group.
clear3d()
par3d(op)
plot3d(rownames(subset(m, cols2==as.character(unique(m$cols2)[7]))), soma=T, col=as.character(unique(m$cols2)[7]))
Eight group.
clear3d()
par3d(op, zoom=0.5743266)
plot3d(rownames(subset(m, cols2==as.character(unique(m$cols2)[8]))), soma=T, col=as.character(unique(m$cols2)[8]))
Last B panel
Plot all target neurons coloured by score for >0, query in red.
clear3d()
plot3d(FCWB)
par3d(op)
plot3d("FruMARCM-M001629_seg001", lwd=2, soma=T, col="red")
plot3d(names(fwdfru_5000[fwdfru_5000>0]), soma=T, col=cols2[fwdfru_5000>0])
For D panels
Plotting of neurons for panels D, explaining the different scores. Comparing normalised and non normalised score for big neuron vs small neuron.
Plot big neuron.
clear3d()
plot3d(FCWB, alpha=0)
op=structure(list(FOV = 0, userMatrix = structure(c(1, 0, 0, 0,
0, -1, 0, 0, 0, 0, -1, 0, 0, 0, 0, 1), .Dim = c(4L, 4L)), scale = c(1,
1, 1), zoom = 0.472224056720734), .Names = c("FOV", "userMatrix",
"scale", "zoom"))
par3d(op)
plot3d("dTdc2MARCM-F000115_seg001")
Plot small neuron in red.
clear3d()
plot3d(FCWB, alpha=0)
op=structure(list(FOV = 0, userMatrix = structure(c(1, 0, 0, 0, 0, -1, 0, 0, 0, 0, -1, 0, 0, 0, 0, 1), .Dim = c(4L, 4L)), scale = c(1, 1, 1), zoom = 0.472224056720734), .Names = c("FOV", "userMatrix", "scale", "zoom"))
par3d(op)
plot3d("FruMARCM-M002574_seg002", col='red')
Plot another small neuron in black.
clear3d()
plot3d(FCWB, alpha=0)
op=structure(list(FOV = 0, userMatrix = structure(c(1, 0, 0, 0, 0, -1, 0, 0, 0, 0, -1, 0, 0, 0, 0, 1), .Dim = c(4L, 4L)), scale = c(1, 1, 1), zoom = 0.472224056720734), .Names = c("FOV", "userMatrix", "scale", "zoom"))
par3d(op)
plot3d("DvGlutMARCM-F004164_seg001", col='black')
Get scores for panels.
fc_nblast("dTdc2MARCM-F000115_seg001", "FruMARCM-M002574_seg002") # big vs small 1## [1] -167916.1fc_nblast("FruMARCM-M002574_seg002","dTdc2MARCM-F000115_seg001") # small 1 vs big## [1] 7987.298fc_nblast("FruMARCM-M002574_seg002","DvGlutMARCM-F004164_seg001") # small 1 vs small 2## [1] 6438.097fc_nblast("DvGlutMARCM-F004164_seg001","FruMARCM-M002574_seg002") # small 2 vs small 1## [1] 5551.83fc_nblast("dTdc2MARCM-F000115_seg001", "FruMARCM-M002574_seg002", norm="mean") # mean score big vs small 1## [1] 0.1359932fc_nblast("DvGlutMARCM-F004164_seg001","FruMARCM-M002574_seg002", norm="mean") # mean score small 2 vs small 1## [1] 0.5788994fc_nblast("dTdc2MARCM-F000115_seg001","dTdc2MARCM-F000115_seg001") #self score big## [1] 400149.2fc_nblast("DvGlutMARCM-F004164_seg001","DvGlutMARCM-F004164_seg001") #self score small 2## [1] 9248.055fc_nblast("FruMARCM-M002574_seg002","FruMARCM-M002574_seg002") #self score small 1## [1] 11548.68Get normalised scores.
fc_nblast("dTdc2MARCM-F000115_seg001", "FruMARCM-M002574_seg002")/fc_nblast("dTdc2MARCM-F000115_seg001","dTdc2MARCM-F000115_seg001") # normalised big vs small 1## [1] -0.4196336fc_nblast("FruMARCM-M002574_seg002","dTdc2MARCM-F000115_seg001")/fc_nblast("FruMARCM-M002574_seg002","FruMARCM-M002574_seg002") # normalised small 1 vs big## [1] 0.6916201fc_nblast("DvGlutMARCM-F004164_seg001","FruMARCM-M002574_seg002")/fc_nblast("DvGlutMARCM-F004164_seg001","DvGlutMARCM-F004164_seg001") # normalised small 2 vs small 1## [1] 0.6003241fc_nblast("FruMARCM-M002574_seg002","DvGlutMARCM-F004164_seg001")/fc_nblast("FruMARCM-M002574_seg002","FruMARCM-M002574_seg002") # normalised small 1 vs small 2## [1] 0.5574747Plots for E panels
Collect the tophit for each neuron (rank 2) from dataframe tophits and plot forward normalised scores as histogram.
tophits2=subset(tophits, query%in%good_images)
normscoresall=mapply(fc_nblast, tophits2[,'target'], tophits2[,'query'], normalisation='mean')
summary(normscoresall)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## -0.8774 0.5195 0.6618 0.5847 0.7178 1.0000quantile(normscoresall, probs=0.99)## 99%
## 0.80191qq=qplot(normscoresall, binwidth=0.01, ylab="Count", xlab="Forward score of top hit")
par(mar=c(0,2,0,0))
qq+theme(axis.text.x=element_text(size=22), axis.text.y=element_text(size=22), axis.title.x=element_text(size=26), axis.title.y=element_text(size=26)) + geom_vline(xintercept = mean(normscoresall), colour="red", linetype=2, size=2) + geom_vline(xintercept = quantile(normscoresall, 0.99), colour="green", linetype=2, size=2)## Warning: position_stack requires constant width: output may be incorrect
Plot of neuron images divided by type: same raw image, same segmentation, same specimen or different specimen
qq2=ggplot(data=subset(top80sym,!duplicated(pair.number) | is.na(pair.number)), aes(x=revscore.norm, y=value.norm)) + geom_point(aes(colour=type), size=10) + labs(x="Reverse score", y="Forward score", colour="Image Type")+theme(axis.text.x=element_text(size=22), axis.text.y=element_text(size=22), axis.title.x=element_text(size=26), axis.title.y=element_text(size=26), legend.text=element_text(size=26), legend.title=element_text(size=26),legend.position=c(0.86, 0.11))
levels(qq2$data$type)<-c("Same segmentation", "Same raw image", "Same specimen", "Different specimen")
plot(qq2)
Inset showing all neuron hits, forward vs reverse normalised scores.
qq3=qplot(data=subset(tophits2, !duplicated(qp)), y=value.norm, x=revscore.norm, xlab="Reverse score", ylab="Forward score")+ geom_vline(xintercept=0.8) +geom_hline(yintercept=0.8)+ theme(panel.background=element_rect(fill="grey80"), panel.grid.minor=element_line(colour='grey80'), axis.text.x=element_text(size=22), axis.text.y=element_text(size=22), axis.title.x=element_text(size=26), axis.title.y=element_text(size=26))
par(mar=c(0,2,0,0))
plot(qq3)