error in hclust function
2
0
Entering edit mode
Alyaa Mahmoud ▴ 440
@alyaa-mahmoud-4670
Last seen 13 months ago
Hi All I am trying to cluster 57 COGs in 24 datasets. I use the following code and run into this error: hc = NULL hc <- hclust(as.dist(1-cor(as.matrix(bb), method="spearman")), method="complete", members=NULL) Error in hclust(as.dist(1 - cor(as.matrix(bb), method = "spearman")), : NA/NaN/Inf in foreign function call (arg 11) In addition: Warning message: In cor(as.matrix(bb), method = "spearman") : the standard deviation is zero hr = NULL hr <- hclust(as.dist(1-cor(t(as.matrix(bb)), method="spearman")), method="complete", members=NULL) I tried to remove any rows that have sd of zero but there was none; ind <- apply(bb, 1, var) == 0 subset <- bb[!ind,] or ind <- apply(bb, 1, sd) == 0 subset <- bb[!ind,] any clue what coule the problem be ? Thanks a lot for your help yours, Alyaa -- Alyaa Mahmoud "Love all, trust a few, do wrong to none"- Shakespeare [[alternative HTML version deleted]]
• 39k views
ADD COMMENT
1
Entering edit mode
Thomas Girke ★ 1.7k
@thomas-girke-993
Last seen 8 weeks ago
United States
You probably forgot to remove the zero variance columns in your matrix. In the step where you are observing the error you are clustering the columns of bb not its rows, since the cor() functions operates on the columns of a matrix not its rows. Running things stepwise might help to pinpoint the problem: ## Sample data bb <- matrix(1:5, 5, 5, dimnames=list(paste("g", 1:5, sep=""), paste("t", 1:5, sep="")), byrow=TRUE)> bb t1 t2 t3 t4 t5 g1 1 2 3 4 5 g2 1 2 3 4 5 g3 1 2 3 4 5 g4 1 2 3 4 5 g5 1 2 3 4 5 > bb t1 t2 t3 t4 t5 g1 1 2 3 4 5 g2 1 2 3 4 5 g3 1 2 3 4 5 g4 1 2 3 4 5 g5 1 2 3 4 5 ## cor() without t() > cor(bb) t1 t2 t3 t4 t5 t1 1 NA NA NA NA t2 NA 1 NA NA NA t3 NA NA 1 NA NA t4 NA NA NA 1 NA t5 NA NA NA NA 1 Warning message: In cor(bb) : the standard deviation is zero ## cor() with t() > cor(t(bb)) g1 g2 g3 g4 g5 g1 1 1 1 1 1 g2 1 1 1 1 1 g3 1 1 1 1 1 g4 1 1 1 1 1 g5 1 1 1 1 1 ## hclust without t() > hc <- hclust(as.dist(1-cor(bb))) Error in hclust(as.dist(1 - cor(bb))) : NA/NaN/Inf in foreign function call (arg 11) In addition: Warning message: In cor(bb) : the standard deviation is zero ## hclust with t() hc <- hclust(as.dist(1-cor(t(bb)))) Thomas On Sat, Jun 02, 2012 at 12:57:46PM +0000, Alyaa Mahmoud wrote: > Hi All > > I am trying to cluster 57 COGs in 24 datasets. I use the following code and > run into this error: > > hc = NULL > hc <- hclust(as.dist(1-cor(as.matrix(bb), method="spearman")), > method="complete", members=NULL) > > Error in hclust(as.dist(1 - cor(as.matrix(bb), method = "spearman")), : > NA/NaN/Inf in foreign function call (arg 11) > In addition: Warning message: > In cor(as.matrix(bb), method = "spearman") : the standard deviation is zero > > hr = NULL > hr <- hclust(as.dist(1-cor(t(as.matrix(bb)), method="spearman")), > method="complete", members=NULL) > > I tried to remove any rows that have sd of zero but there was none; > ind <- apply(bb, 1, var) == 0 > subset <- bb[!ind,] > > or > > ind <- apply(bb, 1, sd) == 0 > subset <- bb[!ind,] > > > any clue what coule the problem be ? > > Thanks a lot for your help > yours, > Alyaa > -- > Alyaa Mahmoud > > "Love all, trust a few, do wrong to none"- Shakespeare > > [[alternative HTML version deleted]] > > _______________________________________________ > Bioconductor mailing list > Bioconductor at r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor
ADD COMMENT
0
Entering edit mode
Hi Thoma Hi James Thanks a lot for your help. @ Thomas: u r right, that did it. @ James: I emailed the data, but thanks a lot its solved already :) Thanks a lot yours, Alyaa On Sat, Jun 2, 2012 at 6:14 PM, Thomas Girke <thomas.girke@ucr.edu> wrote: > You probably forgot to remove the zero variance columns in your matrix. In > the > step where you are observing the error you are clustering the columns of > bb not > its rows, since the cor() functions operates on the columns of a matrix > not its > rows. Running things stepwise might help to pinpoint the problem: > > ## Sample data > bb <- matrix(1:5, 5, 5, dimnames=list(paste("g", 1:5, sep=""), paste("t", > 1:5, sep="")), byrow=TRUE)> bb > t1 t2 t3 t4 t5 > g1 1 2 3 4 5 > g2 1 2 3 4 5 > g3 1 2 3 4 5 > g4 1 2 3 4 5 > g5 1 2 3 4 5 > > > bb > t1 t2 t3 t4 t5 > g1 1 2 3 4 5 > g2 1 2 3 4 5 > g3 1 2 3 4 5 > g4 1 2 3 4 5 > g5 1 2 3 4 5 > > ## cor() without t() > > cor(bb) > t1 t2 t3 t4 t5 > t1 1 NA NA NA NA > t2 NA 1 NA NA NA > t3 NA NA 1 NA NA > t4 NA NA NA 1 NA > t5 NA NA NA NA 1 > Warning message: > In cor(bb) : the standard deviation is zero > > ## cor() with t() > > cor(t(bb)) > g1 g2 g3 g4 g5 > g1 1 1 1 1 1 > g2 1 1 1 1 1 > g3 1 1 1 1 1 > g4 1 1 1 1 1 > g5 1 1 1 1 1 > > ## hclust without t() > > hc <- hclust(as.dist(1-cor(bb))) > Error in hclust(as.dist(1 - cor(bb))) : > NA/NaN/Inf in foreign function call (arg 11) > In addition: Warning message: > In cor(bb) : the standard deviation is zero > > ## hclust with t() > hc <- hclust(as.dist(1-cor(t(bb)))) > > Thomas > > On Sat, Jun 02, 2012 at 12:57:46PM +0000, Alyaa Mahmoud wrote: > > Hi All > > > > I am trying to cluster 57 COGs in 24 datasets. I use the following code > and > > run into this error: > > > > hc = NULL > > hc <- hclust(as.dist(1-cor(as.matrix(bb), method="spearman")), > > method="complete", members=NULL) > > > > Error in hclust(as.dist(1 - cor(as.matrix(bb), method = "spearman")), : > > NA/NaN/Inf in foreign function call (arg 11) > > In addition: Warning message: > > In cor(as.matrix(bb), method = "spearman") : the standard deviation is > zero > > > > hr = NULL > > hr <- hclust(as.dist(1-cor(t(as.matrix(bb)), method="spearman")), > > method="complete", members=NULL) > > > > I tried to remove any rows that have sd of zero but there was none; > > ind <- apply(bb, 1, var) == 0 > > subset <- bb[!ind,] > > > > or > > > > ind <- apply(bb, 1, sd) == 0 > > subset <- bb[!ind,] > > > > > > any clue what coule the problem be ? > > > > Thanks a lot for your help > > yours, > > Alyaa > > -- > > Alyaa Mahmoud > > > > "Love all, trust a few, do wrong to none"- Shakespeare > > > > [[alternative HTML version deleted]] > > > > _______________________________________________ > > Bioconductor mailing list > > Bioconductor@r-project.org > > https://stat.ethz.ch/mailman/listinfo/bioconductor > > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor > -- Alyaa Mahmoud "Love all, trust a few, do wrong to none"- Shakespeare [[alternative HTML version deleted]]
ADD REPLY
0
Entering edit mode
James F. Reid ▴ 610
@james-f-reid-3148
Last seen 6.6 years ago
Hi Alyaa, you probably have missing values (see 'NA/NaN/Inf in foreign function call (arg 11)') in your bb matrix. HTH. J. On 02/06/12 13:57, Alyaa Mahmoud wrote: > Hi All > > I am trying to cluster 57 COGs in 24 datasets. I use the following code and > run into this error: > > hc = NULL > hc<- hclust(as.dist(1-cor(as.matrix(bb), method="spearman")), > method="complete", members=NULL) > > Error in hclust(as.dist(1 - cor(as.matrix(bb), method = "spearman")), : > NA/NaN/Inf in foreign function call (arg 11) > In addition: Warning message: > In cor(as.matrix(bb), method = "spearman") : the standard deviation is zero > > hr = NULL > hr<- hclust(as.dist(1-cor(t(as.matrix(bb)), method="spearman")), > method="complete", members=NULL) > > I tried to remove any rows that have sd of zero but there was none; > ind<- apply(bb, 1, var) == 0 > subset<- bb[!ind,] > > or > > ind<- apply(bb, 1, sd) == 0 > subset<- bb[!ind,] > > > any clue what coule the problem be ? > > Thanks a lot for your help > yours, > Alyaa
ADD COMMENT

Login before adding your answer.

Traffic: 328 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6