Using my interested da peaks in FindMotifs() turns into an error.

[jiangpuxuan]

Here is my code:

de.motif <- rownames(mydata)[1:50]
enriched.motifs <- FindMotifs(
  object = mydata,
  features = de.motif,
)

head(enriched.motifs)

Everything is OK.

And I tried to use my interested da peaks in enriched.motifs :

enriched.motifs <- FindMotifs(
  object = mydata,
  features = the da peaks in some interested cells ,
)

But the error is:

Error in density.default(x = query.feature[[featmatch]], kernel = "gaussian", : 'x' contains missing values

Thank you for your help!

[timoast]

Can you show the DA peaks that cause the error? First few lines and output of dim

[jiangpuxuan]

`de.motif` did not cause the error, but `interested da peaks` did.

[jiangpuxuan]

Can you show the DA peaks that cause the error? First few lines and output of dim

Now I ran chromeVar and found differentially activating motifs , following the https://satijalab.org/signac/articles/motif_vignette.html.

But I still want to know the cause of this error.

[timoast]

How many peaks are in the set of peaks that causes the error?

[jiangpuxuan]

How many peaks are in the set of peaks that causes the error?

About 120.

[pabloswfly]

Curious about the resolution of this issue, I am encountering the same error.
The peak data in my specific case looks like this:

top.da.peak <- rownames(da_peaks[da_peaks$p_val < 0.005,])
head(top.da.peak)
length(top.da.peak)
dim(da_peaks)

----------------------------------------------------------------

'chr3-14451939-14453443''chr2-42903457-42904464''chr18-58364914-58365919''chr4-84417769-84418850''chr1-17602697-17603858''chr1-91585801-91586419'

17242

18728  7

[jiangpuxuan]

Curious about the resolution of this issue, I am encountering the same error. The peak data in my specific case looks like this:

top.da.peak <- rownames(da_peaks[da_peaks$p_val < 0.005,])
head(top.da.peak)
length(top.da.peak)
dim(da_peaks)

----------------------------------------------------------------

'chr3-14451939-14453443''chr2-42903457-42904464''chr18-58364914-58365919''chr4-84417769-84418850''chr1-17602697-17603858''chr1-91585801-91586419'

17242

18728  7

Could you ran chromeVar and found differentially activating motifs?

[pabloswfly]

yes I could, as they are two independent processes.

[jiangpuxuan]

yes I could, as they are two independent processes.

Have you calculated the background in Findmotifs()?

Or I thought that our interested da.peaks are not informative enough so that Findmotifs() get errors.

[pabloswfly]

My error occurs after calculating the background regions, during the matching of GC content:

Selecting background regions to match input sequence characteristics

Matching GC.percent distribution

Error in density.default(x = query.feature[[featmatch]], kernel = "gaussian", : 'x' contains missing values
Traceback:

1. FindMotifs(sub_endocrine, features = top.da.peak)
2. MatchRegionStats(meta.feature = mf.choose, query.feature = mf.query, 
 .     regions = features, n = background, verbose = verbose, ...)
3. density(x = query.feature[[featmatch]], kernel = "gaussian", 
 .     bw = 1)
4. density.default(x = query.feature[[featmatch]], kernel = "gaussian", 
 .     bw = 1)
5. stop("'x' contains missing values")

I had some peaks with zero counts in all cells in my object, but after filtering these rows out, I still get the same error.

[rLannes]

Please, can you tell me How did you filter them out? I am running in the same issue. I would like to give it a try.

[jiangpuxuan]

Please, can you tell me How did you filter them out? I am running in the same issue. I would like to give it a try.

Sorry, I have not solved this. But I recommand you run chromevar pipeline, following the https://satijalab.org/signac/articles/motif_vignette.html.

[rLannes]

I did, and I managed to run chromvar, but the Find Motifs function give me the same error as you. The thing that trouble me is if I use a subset of DE peaks it is fine. That is why I would want to try to subset only the peaks with at least some counts.

enriched.motifs <- FindMotifs(
  object = integrated,
  features = rownames(de)
)
Selecting background regions to match input sequence characteristics
Matching GC.percent distribution
Error in density.default(x = query.feature[[featmatch]], kernel = "gaussian",  : 
  'x' contains missing values

enriched.motifs <- FindMotifs(
  object = integrated,
  features = rownames(de)[1:100]
)
Selecting background regions to match input sequence characteristics
Matching GC.percent distribution
Testing motif enrichment in 100 regions

[rLannes]

I got the solution to my issue. Some peaks in the Differential peak object have an extension "chrX-131703821-131705582.1" removing the .1 solve the issue for me, hope this will help others.

[timoast]

Thanks @rLannes, I've added a check to FindMotifs() for features that are not present in the assay that should resolve these issues.