Closed wb-yun closed 2 years ago
ben-laufer
commented
2 years ago Hi @wb-yun, I haven't yet seen this type of error with DMRichR. Is it possible for you try running this in an earlier version of R, I haven't gotten a chance to test out DMRichR in R 4.1 yet. Also, you can probably drop the cores down to 20. If it happens again, can you give me some more details about your subsampling approach? Since we can take a look at the bsseq object.
wb-yun
commented
2 years ago Thank you for the response. I changed R version from 4.1 to 4.0.5 and dropped the cores down to 20. The process is running. I will share the result.
wb-yun
commented
2 years ago I got error messages.
First error is "DMRs can't be plotted", second error message is "undefined columns selected".
Selecting significant DMRs...
Exporting DMR and background region information...
Saving DMRs/DMRs.bed
Saving DMRs/backgroundRegions.bed
Summary: There are 31 DMRs (42% hypermethylated, 58% hypomethylated) from 482 background regions consisting of 276315 CpGs assayed at 1x coverage
DMR timing...
Saving Rdata...
Annotating DMRs and plotting...
[plotDMRs] Plotting 31 DMRs
[plotDMRs] Smoothing raw methylation values with loess for line plots
ERROR: One (or more) of your DMRs can't be plotted, try again later by manually loading R Data and subsetting sigRegions
[DMRichR] Annotating DMRs with gene symbols 24-09-2021 04:33:47 AM
Preparing HTML report...Annotating 31 regions from GRCh38 with gene symbols
Error in `[.data.frame`(nearestFeatures.df, , cn) :
undefined columns selected
In addition: There were 15 warnings (use warnings() to see them)The process can't reach the previous stop point.
The log of the same step in first try is,
Summary: There are 30 DMRs (40% hypermethylated, 60% hypomethylated) from 482 background regions consisting of 276315 CpGs assayed at 1x coverage
DMR timing...
Saving Rdata...
Annotating DMRs and plotting...
[plotDMRs] Plotting 30 DMRs
[plotDMRs] Smoothing raw methylation values with loess for line plots
ERROR: One (or more) of your DMRs can't be plotted, try again later by manually loading R Data and subsetting sigRegions
[DMRichR] Annotating DMRs with gene symbols 15-09-2021 03:59:19 AM
Preparing HTML report...Annotating 30 regions from GRCh38 with gene symbols
'select()' returned 1:1 mapping between keys and columns
Done
Annotating background regions with gene symbols...
Annotating 482 regions from GRCh38 with gene symbols
'select()' returned 1:many mapping between keys and columns
[DMRichR] Smoothing individual methylation values 15-09-2021 03:59:41 AM
|==============================================================================================================================================| 100%
I just know that the same plotting error was in first try.
Also, I found an additional message in middle of log
Selecting significant blocks...
Exporting block and background information...
Saving Blocks/backgroundBlocks.bed
Blocks timing...
Annotating background blocks with gene symbols...
Annotating 4 regions from GRCh38 with gene symbols
Block analysis has produced an error
Hi @wb-yun, thanks for sharing the logs. The error you're running into this time is related to ensembldb and the workaround for this would be setting EnsDb = FALSE. I added that option in for non-model organisms, so I'd recommend the default UCSC annotations for the human genome. DMRichR will convert the sequence levels automatically for you, so this will still work even if you aligned to GRCh38.
The other two errors you get are more so warnings. Some DMRs can't be plotted, so the program will skip that step instead of stopping and then you can manually plot any of interest that it didn't get to. The other error related to blocks can happen when no significant blocks are detected.
wb-yun
commented
2 years ago Everything is working well without errors. Thank you so much.
One more question, I want to get the input data file for heatmap.pdf file in DMRs directory, how do I have the file?
ben-laufer
commented
2 years ago Glad to hear and you're welcome.
To answer the other question, you'd have to load the RData from its folder and then run the first few lines of code from smoothPheatmap():
bsseq::getMeth(BSseq = bs.filtered.bsseq,
regions = sigRegions,
type = "smooth",
what = "perRegion") %>%
na.omit() %>%
as.matrix()Finally, here's a link to some code showing an example of how to further customize your pheatmap with ComplexHeatmap: https://github.com/ben-laufer/PCB-Placenta-and-Brain/blob/main/ComplexHeatmap.R
Dear author, I got a error message when I tried to use this impressive package for WGBS analysis. I wrote R scripts like this,
DMRichR::DM.R(genome=c("hg38"),testCovariate = "Diagnosis", adjustCovariate = c("Sex","Age"), sexCheck = TRUE, cores = 50, coverage = 1, perGroup = 0.8, minCpGs = 5, maxPerms = 12, maxBlockPerms = 12,cutoff = 0.05,EnsDb = TRUE, cellComposition = FALSE,GOfuncR = TRUE)The error message in R console is,
I think that the error occurred at the step of "ChromHMM and Reference Epigenome Enrichments."
My data are 6 control and 6 cases CpG report files which were generated from subsampling bam files by bismark and
samtools view -bsfunction. However, I don't think that the input files are the problem because Bismark would make intact output files.How to solve this error?
The session info is below,