• Changes
  • Improved efficiency of parameter estimation code
  • Improved efficiency of alignment processing for single-end data
  • Added -XX:+UseSerialGC option to calls to JVM to allow for better parallelization on many-core machines
• Bug fixes
  • Fixed issue caused by read titles containing whitespace. This issue led to read counts of zero for the majority of genes when reads had titles with whitespace separated words.
  • Fixed incorrect determination of read length. This issue led to incorrect computation of likelihoods but only had a minor effect on parameter estimates.
  • Fixed alignment processing for single-end data. This issue led to incorrect computation of likelihoods but only had a minor effect on parameter estimates.

v1.2.0 (2013-10-25)

• New features
  • Single-end data support
  • Added option for number of threads to use
  • Now compatible with Python 2.6
• Changes
  • The reference database produced by psg_prepare_reference.py has changed slightly to fix an issue relating to gene_ids and alignment sorting. Reference databases will need to be rebuilt to use this new version.
  • Changed order of arguments of psg_infer_diff_processing.py for consistency.
  • Now using 1-based gene coordinates in output files.
  • Temporary directory for intermediate files now defaults to Python inferred directory.
• Bug fixes
  • Fixed alignment sorting issue that led to zero-counts for most genes (see reference database change note above)
  • Fixed argument order bug in psg_infer_frequencies.py

v1.1.3 (2013-07-09)

• First software release with paper publication.