This repository has been archived by the owner on Oct 13, 2022. It is now read-only.
0.5.0
Changes from release 0.4.0:
- Performance improvements:
- A new method to compute marker alignments via SeqAn banded alignments is now the default and has improved performance and accuracy. To revert to the old marker alignment algorithm used in previous Shasta releases, use
--Align.alignMethod 0
. - Alignments are now stored, in a highly compressed format, so they don't have to be recomputed when creating marker graph vertices using the disjoint set computation.
- Option
--Reads.noCache
can be used to bypass the Linux cache when loading reads. This can improve performance in some situations. - Several other performance improvements, including upgrading to a new, faster release of the Spoa library. As a combined result of these and the above, a human genome assembly at coverage 60x now takes about 3 hours on a
x1.32xlarge
AWS instance.
- A new method to compute marker alignments via SeqAn banded alignments is now the default and has improved performance and accuracy. To revert to the old marker alignment algorithm used in previous Shasta releases, use
- New functionality:
- Option
--Assembly.detangle
performs basic detangling in the assembly graph and can improve assembly contiguity. - Option
--Assembly.writeReadsByAssembledSegment
can be used to write a csv file containing the reads and orientations that were used to assemble each segment. - New options to generate the k-mers to be used as markers.
- Usability improvements in the Shasta http server, including improved display of a read and its markers.
- Option
- New configuration files and Bayesian model for Oxford Nanopore reads created by the Guppy 3.6.0 base caller.
- Fixed a long-standing bug in the computation of CIGAR strings in GFA output.
- Platform changes:
- Shasta can now be built on Ubuntu 20.04, in addition to 16.04 and 18.04. As for previous releases, the static executable built on Ubuntu continues to run on most current 64-bit Linux platforms, has no dependencies, and requires no installation.
- Support for MacOS 10.15 Catalina (both build and run).
- GPU support was removed. Because of the above improvements in performance, the GPU code was no longer providing performance benefits.
- Many documentation improvements which make it easier to locate the desired information.
- Code reorganization and cleanup.
This release is not compatible with previous releases. There were incompatible changes in some command line option names, the binary formats used, and the Python API. You cannot use release 0.5.0 for postprocessing of an assembly done using a previous release. If you plan to build the code yourself and you have done so for previous Shasta releases, make sure to rerun InstallPrerequisites-Ubuntu.sh
or InstallPrerequisites-macOS.sh
in order to get updated prerequisites.