Repository navigation
0.111.0
Major changes are:
Support for joint analysis has been added using the -iso_joint option.
Isoquant has stricter requirements for the detection of novel isoforms than for known ones. This can cause novel isoforms to be missed in some samples. Joint analysis works by first analysing all samples separately, and than using the novel isoforms found in any of the samples as "known/reference" ones in a reanalysis (still per sample), causing them to be more likely found (and counted) even in these lower evidence samples. If you have many samples, the number of artefactual "novel" transcripts can become too large to use as a reference (performance wise). With the option -iso_joint_min only novel transcripts found in at least samples are used as a reference in stage2, reducing the number of artefacts used drastically.
The command sc_demultiplex was added for demultiplexing singlecell results into different samples, based on a demultiplexing file assigning each cell to sample. (Such a file can be generated based on genomic variants or feature barcode sequencing)
Some improvements in sc barcode detection were made:
- find_barcodes handles the presence of multiple adapter matches in a read better (mapquality filter, location)
- Support for v4 and 5 prime 10x barcodes has been added
An alternative cluster distribution method that limits the number of concurrently submitted jobs was added: When using -d slurm, all jobs will be submitted immediately (in a manner jobs will only be run when jobs they are dependent on are finished). This can (for large analyses) submit thousands of jobs, causing problemson clusters that limit the number of submitted jobs. You can now use the options -d -dsubmit slurm to run (distributed) on a slurm cluster, while limiting the number of concurrently submitted jobs to . The disadvantage of this method is that the submitting command has to keep running until the entire command/pipeline is finished (as it is managing the jobs partly by itself).
A custom set of regions for distribution can be added to a genome reference, and used by the option -distrreg g. This allows for optimization of parallel analysis by splitting up regions that take a longer time to process (due to e.g. size of gene richmess).
Isoquant has been updated to version 3.6.3 (issue #20)
ubams are now supported as source data.
A basic singlecell report is generated even if some data is missing
The following issues are also solved in this release:
Full Changelog: 0.110.0...0.111.0