v0.10.0 Isoform-level EM and bias correction methods

This is a major update, strongly suggested for all users.

NOTE: Whippet v0.9 Indexes are not compatible with Whippet v0.10!

Whats new in the release?

• --ebi and --url features are back by popular demand!
• 5' Sequence bias correction method by Hansen et al. NAR 2010 for both TPM and PSI calculations
• Fragment GC-content bias correction for PSI calculations. Background GC-content is built into the Whippet Index for each transcript, expression weighted after isoform-level EM and then used to correct edge counts.
• General performance increase: Run-time decrease for EM stage which now scales as a function of transcriptome complexity rather than read-depth! This is due to multi-mapping read storage into alignment path hashed equivalence classes.
• Performance-related bugfix for event-level MLE stage which affected the run-time of certain high-depth datasets.
• More accurate TPM quantification, producing TPM values for all annotated isoforms normalized by exact isoform length (vs. gene-level splice graph average before). New output file formats are .isoform.tpm.gz and .gene.tpm.gz.
• Multi-gene mapping reads whose edges don't fit any annotated isoform are assigned after EM based on gene-level proportions.
• Increased default iterations for EM which now completely converge in a few seconds, plus short-circuit of equivalence classes (isoform-level and multi-gene level) whose proportions have already converged.
• Bugfix in how multi-mapping paired reads were previously double counted during gene-expression EM.

v0.9.0 Julia v0.6 Compatibility Release

Whats new in the release?

• Julia v0.6 compatible, and dropped support for Julia v0.5
• Suggested installation now through Julia package manager
• MIT Licensed
• Temporarily deprecated --ebi/--url support until dependent packages update.

v0.8 Full Output Node Release

This release v0.8 produces .psi.gz output which is incompatible with previous versions of Whippet. Some notable changes are:

• All nodes (even NA type) are outputted in .psi.gz file for downstream plotting.
• Maximum iteration number to restrict runaway reduce-graph behavior, and alternate version for constitutive inclusion/skipping.
• Minimum weight threshold for skipping reads (default 1%)

v0.7.3 Bugfix & Edges Column Output

This release:

• Fixes a critical "graph reduction" bug #46 introduced when adding constitutive node output in v0.7.1. (the bug basically made all psi values in complex events wildly wrong.
• Adds an additional column to .psi.gz output containing all edges and their read counts

v0.7.2 Bugfix and TSL-dependent index warning

This release:

• Provides a bugfix for NaN TpM output with some indexes
• Warns users when they use a GTF file containing TSL2+ transcript annotations
• Provides users with an option to suppress TSL2+ transcript annotations --suppress-low-tsl

v0.7.1 Constitutive Splicing Output

• This release includes a major change to constitutively included or excluded nodes, replacing trailing NAs with data.
• Bugfix for whippet-delta.jl probabilities
• Removed -m option for whippet-delta.jl and added --seed option and default
• Slightly improved docs for whippet-delta.jl

v0.7 Output Change Release

This release has a number of changes:

• Output format change for Inc/ExcPaths
• Bugfix for AbstractString usage by ArgParse
• Change in EM length usage for GE
• Code Cleanup
• Updated Docs

v0.6.3 Performance adjustment

v0.6.3 contains a performance improvement in alignment, and restores a few attempted performance tweaks that were counter productive.

v0.6.2 Second Bugfix Release

• This release fixes the isvalid alignment issue for paired end reads, both fwd and rev multimapping reads must have valid alignments to be counted.
• This release renames --curl flag for --urlsince curl library is no longer used.

v0.6.1 Bugfix Release

Fixes some alignment isvalid checks.