[adegenet-forum] Plot supplementary individuals DAPC

[Thibaut Jombart]

Hi there,

it could be 2 things: a normal behaviour, or a bug. For the first, the
result could make sense: individuals will be assigned to their closest
group, even if they fall outside the group's cloud of point. So the result
in itself may not be too surprising.

The fact that the result changed with the version of adegenet is more
concerning at it suggests a bug. However, the DAPC itself hasn't changed
much. Can you provide a reproducible example showing the change?

Best
Thibaut


--
Dr Thibaut Jombart
Lecturer, Department of Infectious Disease Epidemiology
Imperial College London
https://sites.google.com/site/thibautjombart/
https://github.com/thibautjombart
Twitter: @TeebzR <https://twitter.com/TeebzR>

On 1 September 2016 at 14:07, vlabeyrie <vanesse.labeyrie at cirad.fr> wrote:

> Dear adegenet users,
>
> I have a problem when plotting supplementary individuals on DAPC
> scatterplot: while they are clearly assigned to a group (with a very high
> probability), they are displayed outside of this group on the scatterplot.
>
>
> I defined two datasets: 1 to perform DAPC and another of supplementary
> individuals
>
> x.sup_80<-Ge_atp_gcp_80[c(1:nrow(Ge_atp at tab)),] # supplementary
> individuals
>
> x_80<-Ge_atp_gcp_80[-c(1:nrow(Ge_atp at tab)),] # Individuals on which
> performing DAPC
>
> Then I performed DAPC on X_80, specifying a-priori groups
>
> dapc_GCP_14ssr_STRk5_b<- dapc(x_80, pop(x_80), n.pca=30,n.da=4) #perform
> DAPC
>
> I assigned supplementary individuals to DAPC groups
>
> predict_atp_strk5<-predict.dapc(dapc_GCP_14ssr_STRk5_b,newdata=x.sup_80
>
> The predicted group memberships of supplementary individuals based on DAPC
> results is high, so I expect that supplementary individuals would be
> located in the DAPC groups  ...
>
> But it is not the case, as on the scatterplot, supplementary individuals
> mostly appear outside from the groups to which they are assigned  !!
>
> col<-c("#F8766D", "#A3A500", "#00BF7D", "#00B0F6", "#E76BF3") # colors for
> DAPC individuals
> colb<-c("darkblue","dodgerblue","darkorange2","red","gold","grey") #
> colors for supplementary  individuals
>
> #axes 1 and 2
> col.points_80<-transp(col[as.integer(pop(x_80))],.2) # define the color
> of DAPC individuals as transparent
> scatter(dapc_GCP_14ssr_STRk5_b,col=col,bg="white",scree.da=
> 0,pch="",cstar=0,clab=0,xlim=c(-10,10),legend=F)#
> par(xpd=TRUE)
> points(dapc_GCP_14ssr_STRk5_b$ind.coord[,1], dapc_GCP_14ssr_STRk5_b$ind.
> coord[,2],pch=20,col=col.points_80,cex=1) ## scatter DAPC groups
>
> col.sup_80<-colb[as.integer(pop(x.sup_80))] ## Define supplementary
> individuals color
> points(predict_atp_strk5$ind.scores[,1], predict_atp_strk5$ind.scores[,
> 2],pch=8,col=transp(col.sup_80,<
> span class="number" style="box-sizing: border-box; color: rgb(0, 153,
> 153);">.7),cex=1) # plot supplementary individuals
>
>
> With a previous version of adegenet, this problem did not appear as
> supplementary individuals were located within the DAPC groups, and this
> problem appeared while running my script with the new adegenet version ...
>
> Does someone have an idea of what the problem is ? I can provide the
> picture of the scatterplot, full script and data if needed
>
>
> Thank you for your help !
>
> --
> Vanesse Labeyrie
>
>
>
>
> _______________________________________________
> adegenet-forum mailing list
> adegenet-forum at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/
> listinfo/adegenet-forum
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/adegenet-forum/attachments/20160901/576324aa/attachment.html>