Multiplex_Major_Patch #452
Conversation
This reverts commit 9b865dfa334e22774a9e51fd829afd23b7e27181.
| combinedTranscripts <- as_tibble(rbindlist(list(combinedSplicedTranscripts, | ||
| combinedUnsplicedTranscripts), fill = TRUE)) | ||
| return(combinedTranscripts) | ||
| return(combinedSplicedTranscripts) |
There was a problem hiding this comment.
Hi @andredsim
where is the commented code moved to? does this mean there is no unsplicedNew transcript models?
There was a problem hiding this comment.
This is an outdated version, just double check you have pulled the latest version of this branch, as the commented code is uncommented in it. Now the unsplicedNew transcript models only appear if the user provides min.txScore.singleExon < 1. Which is essentially the same as before as we didn't output novel single exon transcripts by default.
| if(length(annotationGrangesList)>0){ #only recommend an NDR if its possible to calculate an NDR | ||
| NDR = recommendNDR(rowDataCombined, baselineFDR, NDR, defaultModels, verbose) | ||
| } else { | ||
| if(is.null(NDR)) NDR = 0.5 | ||
| } | ||
| filterSet = (rowDataCombined$NDR <= NDR) | ||
| filterSet = (rowDataCombined$NDR <= NDR | rowDataCombined$readClassType == "equal:compatible") |
There was a problem hiding this comment.
why the equal read class needed here? would they just be the annotated transcripts?
small change
| @@ -136,24 +136,48 @@ | |||
| #' genome = fa.file, discovery = TRUE, quant = TRUE) | |||
| #' @export | |||
| bambu <- function(reads, annotations = NULL, genome = NULL, NDR = NULL, | |||
| opt.discovery = NULL, opt.em = NULL, rcOutDir = NULL, discovery = TRUE, | |||
| quant = TRUE, stranded = FALSE, ncore = 1, yieldSize = NULL, | |||
| mode = NULL, opt.discovery = NULL, opt.em = NULL, rcOutDir = NULL, discovery = TRUE, | |||
There was a problem hiding this comment.
the mode parameter is very helpful, would be great to have pre-defined parameters for each mode documented in the readme or documentation
| lowMemory = TRUE | ||
| } | ||
| if(mode == "multiplexed"){ | ||
| if(is.null(demultiplex)) demultiplex = TRUE |
There was a problem hiding this comment.
| if(is.null(demultiplex)) demultiplex = TRUE | |
| if(is.null(demultiplex)) demultiplexed = TRUE |
There was a problem hiding this comment.
is it better to default it to FALSE when mode is "multiplexed"?
then give option for providing demultiplexed bam file
| fusionMode = FALSE, verbose = FALSE) { | ||
| fusionMode = FALSE, verbose = FALSE, demultiplexed = FALSE, spatial = NULL, quantData = NULL, | ||
| sampleNames = NULL, cleanReads = FALSE, dedupUMI = FALSE, barcodesToFilter = NULL, clusters = NULL) { | ||
| message(paste0("Running Bambu-v", "3.3.0")) |
There was a problem hiding this comment.
| message(paste0("Running Bambu-v", "3.3.0")) | |
| message(packageVersion("bambu")) |
There was a problem hiding this comment.
not sure why this is needed though
Work in progress. Do not use or run
In this PR transcripts that are normally filtered out including subset transcripts and those above the NDR threshold are placed into the metadata of the annotations in $subsetTranscripts and $lowConfidenceTranscripts respectively.
This means that if users ran bambu with the wrong NDR setting, and do not want to run discovery again, they can get the missing transcripts from the metadata.
To facilitate this this PR adds the external setNDR function which takes the extendedAnnotations and an NDR value and will switch novel bambu annotations between the main annotations and the low confidence annotations based on the threshold provided. If no threshold is provided, setNDR will recommend an NDR with the same method used during transcript discovery.
In order for this to work for annotations that have already been saved to a gtf file, bambu now outputs the NDR, txScore and txScore.noFit as attributes to the gtf file and these are also read in with prepareAnnotations.
Important to note that if annotations are written with an NDR threshold of <1, these low confidence transcripts will be missed.
Added setNDR as part of quant, which means that users can provide their extendedAnnotations alongside an NDR threshold when running bambu and it will automatically adjust the NDR used for quant. This means users do not need to manually filter the NDR value themselves.
NDR and other stats are now copied over to equal transcripts even if above the NDR threshold (previously only happened for those below the NDR threshold)
Minor change:
Warnings will no longer occur if there are seqlevels in the readGrgList that are not in the annotations or genome. This was done by setting seqlevels of the reads to only those in the reads. Warning was constantly occuring because all the scaffolds used in alignment were in the bam files, even if no reads from these scaffolds existed.
Todo - Unit Tests, Update bambu documentation to include setNDR