BiocParallel error: cannot open the connection, how do I fix it? - r

Question

I'm trying to use the package bambu to quantify gene counts from bam files. I am using my university's HPC, so I have written an R script and a batch submission file to launch it.
When the script gets to the point of running the bambu function, it gives the following error:
Start generating read class files
| | 0%[W::hts_idx_load2] The index file is older than the data file: ./results/minimap2/KD_R1.sorted.bam.bai
[W::hts_idx_load2] The index file is older than the data file: ./results/minimap2/KD_R3.sorted.bam.bai
[W::hts_idx_load2] The index file is older than the data file: ./results/minimap2/WT_R1.sorted.bam.bai
[W::hts_idx_load2] The index file is older than the data file: ./results/minimap2/WT_R2.sorted.bam.bai
|================== | 25%
Error: BiocParallel errors
element index: 1, 2, 3
first error: cannot open the connection
In addition: Warning message:
stop worker failed:
attempt to select less than one element in OneIndex
Execution halted
So it looks like BiocParallel isn't happy and cannot open a certain connection, but I'm not sure how to fix this?
This is my R script:
#Bambu R script
#load libraries
library(Rsamtools)
library(bambu)
#Creating files
bamFiles<- Rsamtools::BamFileList(c("./results/minimap2/KD_R1.sorted.bam","./results/minimap2/KD_R2.sorted.bam","./results/minimap2/KD_R3.sorted.bam","./results/minimap2/WT_R1.sorted.bam","./results/minimap2/WT_R2.sorted.bam","./results/minimap2/WT_R3.sorted.bam"))
annotation<-prepareAnnotations("./ref_data/Homo_sapiens.GRCh38.104.chr.gtf")
fa.file<-"./ref_data/Homo_sapiens.GRCh38.dna.primary_assembly.fa"
#Running bambu
se<- bambu(reads=bamFiles, annotations=annotation, genome=fa.file,ncore=4)
se
seGene<- transcriptToGeneExpression(se)
#Saving files
save.file<-tempfile(fileext=".gtf")
writeToGTF(rowRanges(se),file=save.file)
save.dir <- tempdir()
writeBambuOutput(se,path=save.fir,prefix="Nanopore_")
writeBambuOutput(seGene,path=save.fir,prefix="Nanopore_")
If you have any ideas on why this happens it would be so helpful! Thank you

Answer 1

I think that #Chris has a good point. Under the hood it seems likely that bambu is running htslib based on those warnings. While they may indeed only be warnings, I would like to know what the results would look like if you ran this interactively.
This question is hard to answer right now as it's missing some information (what do the files look like, a minimal reproducible example, etc.). But in the meantime here are some possibly useful questions for figuring it out:
what does bamFiles look like? Does it have the right number of read records? Do all of those files have nonzero read records? Are any suspiciously small?
What are the timestamps on the bai vs bam files (e.g. ls -lh /results/minimap2/)? Are they about what you'd expect or is it wonky? Are any of them (say, ./results/minimap2/WT_R2.sorted.bam.bai) weirdly small?
What happens when you run it interactively? Where does it fail? You say it's at the bambu() call, but how do you know that?
What happens when you run bambu() with ncores=1?
It seems very likely that this is due to a problem with the files, and it is only at the biocParallel step that the error is bubbling up to the top. Many utilities have an annoying habit of being happy to accept an empty file, only to fail confusingly without informative error messages when asked to do something with the empty file.
You might also consider raising an issue with the developers.
(why the warning is only possibly a problem: The index file sometimes has a timestamp like that for very small alignment files which are generated and indexed programmatically, where the indexing step is near-instantaneous.)