v0.7.0

Install oarfish 0.7.0

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/COMBINE-lab/oarfish/releases/download/v0.7.0/oarfish-installer.sh | sh

Key changes in oarfish 0.7.0

This version of oarfish brings with it several substantive changes / improvements. First, several dependencies have been upgraded to their latest major or minor versions. This includes a move to the long-awated rand 0.9. This version also adopts the latest version of minimap2-rs.

Accompanying this change is improved memory-safety and multithreaded ability. However, without any additional tuning, we noticed substantial increases in the memory requirements in raw-mapping based mode as the number of threads grows large. This has been addressed in this release, and is the result of the rather large per-thread default maximum memory allocation allowed by minimmap2. A corresponding command line option has been added to allow controlling the total thread-local buffer size, with a default of 1G (to be divided evenly over all threads). This can be controlled by the --thread-buff-size command line option.

Additionally, the coverage model has been updated and improved. We thank an anonymous reviewer for pointing out an edge case with the previous coverage model (transcripts with only 2 bins), as well as hinting at the, perhaps, unnecessary complexity of the model. In v0.7.0 we have adopted a new coverage model based on a probability obtained by passing the (mean-scaled) deviation of each bin from the expected coverage through a logistic function. The default growth rate or (k) parameter of this logistic function is set to 2, but it can be modified by the user via the --growth-rate command line option. This new coverage model has the benefit of avoiding the degenerate 2-bin case, of being conceptually simpler, and of being numerically more robust. In our testing so far, we have found that it performs similarly to, though often better than, the previous modified Binomial model. Nonetheless, since the model itself is changing, we recommend that within one analysis you ensure that all samples processed using the coverage model are processed by the same version of oarfish to avoid any potential batch effects. Samples processed without the coverage model should not exhibit any differences from the previous version.

Download oarfish 0.7.0

File	Platform	Checksum
oarfish-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
oarfish-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
oarfish-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

v0.6.5

Install oarfish 0.6.5

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/COMBINE-lab/oarfish/releases/download/v0.6.5/oarfish-installer.sh | sh

Download oarfish 0.6.5

File	Platform	Checksum
oarfish-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
oarfish-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
oarfish-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

v0.6.4

Install oarfish 0.6.4

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/COMBINE-lab/oarfish/releases/download/v0.6.4/oarfish-installer.sh | sh

Download oarfish 0.6.4

File	Platform	Checksum
oarfish-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
oarfish-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
oarfish-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

v0.6.3

Install oarfish 0.6.3

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/COMBINE-lab/oarfish/releases/download/v0.6.3/oarfish-installer.sh | sh

Download oarfish 0.6.3

File	Platform	Checksum
oarfish-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
oarfish-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
oarfish-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

v0.6.2

Changes

This version fixes a potential bug in raw-read based mode (which was introduced in v0.6.1) when using the coverage model. A shift of 1 base (due to a 0-1 coordinate transform) could cause a mapped position to extend past the end of a transcript and prevent running with the coverage model enabled on some data. This version addresses that bug.

This also bumps the versions of some dependencies.

Install oarfish 0.6.2

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/COMBINE-lab/oarfish/releases/download/v0.6.2/oarfish-installer.sh | sh

Download oarfish 0.6.2

File	Platform	Checksum
oarfish-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
oarfish-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
oarfish-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

v0.6.1

Install oarfish 0.6.1

Release 0.6.1 adds two substantial new features to oarfish. Both are currently considered to be in beta, but have been tested and appear to work well. However, as always, please report on GitHub if you run into any unexpected issues.

New Features

read-based mode

The first new feature is the capability to quantify directly from a reference and sequencing reads (i.e. bypassing the need to first generate a bam file). To perform this task, oarfish takes advantage of the excellent minimap2-rs library. When run in this mode, oarfish will index the reference transcriptome and then align the reads against this index (you can also pass a pre-built index in place of the reference if you have one). This allows one to align and quantify transcript abundance in a single-command, avoids the need to generate a potentially large intermediate bam file, and also minimizes the possibility of using wrong or sub-optimal mapping parameters. When using this mode, you must also provide the --seq-tech argument; see the documentation for more details.

single-cell quantification

This release incorporates the first single-cell quantification capabilities. Given a bam file, collated by cell barcode and with already (UMI) deduplicated reads, this mode, enabled with the --single-cell flag, will allow oarfish to produce a single-cell quantification matrix. Currently, this mode cannot be used with read-based mode, and the input bam file should be properly formatted for this purpose (i.e. all alignment records for the same cell barcode should be adjacent in the bam file, and a count will be obtained for each read record, so UMI de-duplication should have been performed if those are the counts you want). In the future, counting UMIs directly may be supported, and some of these other restrictions may be lifted.

Improvements

It is now easier (trivial) to override parameters present in a filter-group. That is, if you choose an existing filter-group, and then additionally provide specific values for any of the filters, the provided individual values will override the default values of the filter group. This behavior mimics the "preset"/override behavior of other tools like minimap2 and should increase the usability of the filter groups. It is particularly helpful for overriding directional filters (i.e. to keep only alignments in a certain orientation).

Other various enhancements, including a nice spinner to show progress when processing a bam file or raw reads, has been added.

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/COMBINE-lab/oarfish/releases/download/v0.6.1/oarfish-installer.sh | sh

Download oarfish 0.6.1

File	Platform	Checksum
oarfish-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
oarfish-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
oarfish-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

v0.5.1

New features

oarfish now outputs a file of the form <output_stem>.ambig_info.tsv. This file has a number of rows equal to the number of transcripts in the quantified transcriptome (plus a header). It records, for each transcript (in the same order in which they appear in <output_stem>.quant) the number of unique, ambiguous, and total (unique + ambiguous) reads aligning to each transcript.

Improvements

oarfish v0.5.1 will dedicate > 1 thread to bam file chunk decompression if >2 threads are passed on the command line.

Install oarfish 0.5.1

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/COMBINE-lab/oarfish/releases/download/v0.5.1/oarfish-installer.sh | sh

Download oarfish 0.5.1

File	Platform	Checksum
oarfish-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
oarfish-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
oarfish-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

v0.5.0

Install oarfish 0.5.0

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/COMBINE-lab/oarfish/releases/download/v0.5.0/oarfish-installer.sh | sh

Download oarfish 0.5.0

File	Platform	Checksum
oarfish-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
oarfish-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
oarfish-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

v0.4.0

Install oarfish 0.4.0

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/COMBINE-lab/oarfish/releases/download/v0.4.0/oarfish-installer.sh | sh

Download oarfish 0.4.0

File	Platform	Checksum
oarfish-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
oarfish-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
oarfish-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

v0.3.1

Install oarfish 0.3.1

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/COMBINE-lab/oarfish/releases/download/v0.3.1/oarfish-installer.sh | sh

Download oarfish 0.3.1

File	Platform	Checksum
oarfish-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
oarfish-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
oarfish-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum