HATCHet v0.4.14

This release of HATCHet adds a check command that will check your environment for runtime dependencies that HATCHet relies on (like a useful LP solver like Gurobi). Currently this command behavior is identical to the check-solver command. The check-solver command is still supported for now, but users are encouraged to use the check command instead. This will allow us to add more comprehensive runtime checks in the future.

HATCHet v0.4.12

Fixed a bug in the count_alleles step which would have caused failure to generate temporary files, if an output directory is explicitly specified.

HATCHet v0.4.10

This release of HATCHet fixes a bug where multiple parallel runs of the count_alleles step for the same chromosome (e.g. for multiple patients) would have created temporary files that would clash with each other. This temporary file creation step is now protected by using facilities from the tempfile module.

HATCHet v0.4.9

This HATCHet release fixes a bug:

• End users who built HATCHet from source using the environment variable HATCHET_BUILD_NOEXT set to 1 (i.e. not building the fast C++ solver extension of HATCHet that requires Gurobi to be pre-installed, but instead relying on pyomo-supported solvers that HATCHet also supports), would previously have failed to use the compute_cn step of HATCHet. This is fixed now.

There are two enhancements:

• The compute_cn step, when using pyomo solvers, will now output the number of variables and constraints, helpful to determine the size of the problem that this factorization step is trying to solve.
• Faster and more atomic unit tests to aid development efforts.

HATCHet 0.4.7

This version of HATCHet fixes 2 bugs:

• Running the HATCHet compute-cn step (using the default cpp solver) on non-English locales could have caused a parsing error, either by the ProgressBar class or while parsing the final optimized value. See issue 86.
• Running anything other than the default cpp solver in non-ILP mode would have caused a crash due to the cd.run returning a 6-tuple instead of a 4-tuple as was the case in earlier versions. This is fixed now.

In addition, the timelimit parameter passed from the command line to the compute-cn step is now respected when using any pyomo-supported solvers. See discussion here.

HATCHet 0.4.6

This version of HATCHet fixes a bug (experienced when using a pyomo solver through a HATCHET_COMPUTE_CN_SOLVER environment variable). When parsing clonal CN strings (28:1:1,32:2:1 etc.) to solve, the python implementation of the solver insisted that ALL clonal segments have their total copy numbers different, which is a bug. In reality, only the first two (used for scaling) need to have their total copy numbers different (as was the case when using the default CPP implementation of the solver). This has now been fixed.

HATCHet 0.4.5

This version of HATCHet supports a compute_cn.timelimit configuration variable in hatchet.ini, to set the time limit (in seconds) for each ILP step of supported pyomo solvers. Note that this ini variable already existed for the built-in CPP solver, but is now also used in case of pyomo solvers.

This configuration variable can also be specified as an environment variable (as can any ini variable for that matter), in this case, HATCHET_COMPUTE_CN_TIMELIMIT. If left unspecified, a default time limit of 100 seconds is used internally by HATCHet.

HATCHet 0.4.4

This version of HATCHet has a better error message when we encounter a GRBException while running the legacy CPP solver compiled against Gurobi, suggesting that the user run the hatchet check-solver command and get that working first.

HATCHet 0.4.3

This version of HATCHet introduces new command check-solver. check-solver provides a fast way to run the HATCHet solver (to determine tumor-family specific mixing proportions) on a small set of pre-packaged data files, and should help debug issues with solver license etc. faster.

HATCHet 0.4.2

This version of HATCHet supports multiple ILP solvers (i.e. any solver supported by the pyomo module), not just Gurobi (though Gurobi is still supported and recommended for fastest factorization to determine allele-specific copy numbers).

To use this feature, set HATCHET_COMPUTE_CN_SOLVER environment variable to a pyomo-supported solver. For example export HATCHET_COMPUTE_CN_SOLVER=cbc is one setting that has been extensively tested.