Hi, I am using the Kallisto-Sleuth pipeline for differential expression of RNA-seq and I am having a problem with performing 'sleuth_prep' in Sleuth. It's my first time to use R package Sleuth, I dont know how solve the problem, any help would be fantastic for me!
t2g <- biomaRt::getBM(attributes = c("ensembl_transcript_id", "ensembl_gene_id",
"external_gene_name"), mart = mart)
t2g <- dplyr::rename(t2g, target_id = ensembl_transcript_id,
ens_gene = ensembl_gene_id, ext_gene = external_gene_name)
library(sleuth)
so <- sleuth_prep(s2c, ~ condition, target_mapping = t2g, extra_bootstrap_summary = TRUE)
reading in kallisto results
dropping unused factor levels
.....
normalizing est_counts
26768 targets passed the filter
normalizing tpm
merging in metadata
Error in H5Fopen(file, "H5F_ACC_RDONLY", native = native) :
HDF5. File accessibilty. Unable to open file.
Warning message:
In check_target_mapping(tmp_names, target_mapping, !is.null(aggregation_column)) :
intersection between target_id from kallisto runs and the target_mapping is empty. attempted to fix problem by removing .N from target_id, then merging back into target_mapping. please check obj$target_mapping to ensure this new mapping is correct.
•
written
by
ladavion •
Just use the path of folder with h5 file, and use the kallisto output file without modification of the file name, it works well
Can you please be more specific?; and show exact commands, where necessary? Some of the issues to which I linked do not relate to Sleuth, so, may not actually be the same error.
Log in to answer this question.
I don't think that there has ever been a clear solution to this problem. It has been documented already, though. - see my comment on Bioconductor, with links to potential solutions: https://support.bioconductor.org/p/130419/#130424
@Kevin Blighe Thanks for your help
I am having the same problem. I have tried to figure out how to implement the workarounds, but I am creating the files running jobs on a cluster and 1. the versions listed aren't available and 2. I am just not quite at the level where I can create environments when I run a job, unless that just means setting the modules, in which case I could but the versions don't match the suggested ones...
Hey Jocelyn, can you possibly elaborate on the pipeline that you are trying to follow? - give program and OS versions, where possible.
Hello, I was trying to use the pipeline as described in the sleuth documentation for going kallisto->sleuth. I have found the same problem documented a few times re: h5 files in the same process. Right now, I cannot locate the tutorial I was looking to go through it, but as soon as it came to loading the h5 file, I had an issue. Unfortunately, I did not save the code I had used. Generally I run everything in an R Markdown document, but for whatever reason I neglected to that evening and shut RStudio, losing what I was running on the commend line. I believe it used the cowplot data and included a portion re: downloading the information to go from ensembl transcript ids -> gene names. Based on what I read re: this exact error, it was not the code I was running but the system I was running kalliso on when the h5 files were created. I would assume I was running things on a linux based cluster, but I did not pay that much attention to the node that the jobs were run on. It isn't a thing I knew had any reason to take note of at the time. My own computer is running macOS 10.15.5. I do know that the cluster does not have the latest version of kallisto (I think it is one or two behind whatever environment was set up as a potential solution on one of the git issue threads) and I do not think the same version of the h5 creating library/package/?, but I didn't need to load that module when I used kallisto. I did not know which would work correctly give the version of kallisto I was running (kallisto/0.45.0)
Yep, that's the conclusion that I got, too. The error has been reported a few times, as you mention, but no definitive solution or exact diagnosis.
This was my solution, which Ive now had to use twice on 2 different systems:
I did have an exchange with the rhdf5 R package developer, and he managed to solve an issue causing this error for one user, but it wasn't a panacea / global cure