minor pymc/pytensor version upgrades, fix 2-interval edge case, updat…

…e some theano docs

scripts/gatkcondaenv.yml.template

@@ -22,12 +22,12 @@ dependencies:
 - conda-forge::python=3.10.13         # do not update without good reason
 - conda-forge:pip=23.3.1
 - conda-forge:blas=1.0=mkl            # our official environment uses MKL versions of various packages; if other versions are desired, users should edit this YML accordingly
-- conda-forge::numpy=1.26.2           # MKL
-- conda-forge::pymc=5.10.0            # do not update without good reason, this may break gcnvkernel
-- conda-forge::pytensor=2.18.1        # do not update without good reason, this may break gcnvkernel
+- conda-forge::numpy=1.26.2
+- conda-forge::pymc=5.10.1
+- conda-forge::pytensor=2.18.3
 - conda-forge::scipy=1.11.4
 - conda-forge::h5py=3.10.0
-- conda-forge::pytorch=2.1.0=*mkl*100 # CPU MKL
+- conda-forge::pytorch=2.1.0=*mkl*100
 - conda-forge::scikit-learn=1.3.2
 - conda-forge::matplotlib=3.8.2
 - conda-forge::pandas=2.1.3

src/main/java/org/broadinstitute/hellbender/tools/copynumber/DetermineGermlineContigPloidy.java

@@ -67,12 +67,12 @@
  * <p>OpenMP and MKL parallelism can be controlled by setting the <code>OMP_NUM_THREADS</code> and <code>MKL_NUM_THREADS</code>
  * environment variables, respectively.</p>
  *
- * <p>Advanced users may wish to set the <code>THEANO_FLAGS</code> environment variable to override the GATK theano
+ * <p>Advanced users may wish to set the <code>PYTENSOR_FLAGS</code> environment variable to override the GATK PyTensor
  * configuration. For example, by running
- * <code>THEANO_FLAGS="base_compiledir=PATH/TO/BASE_COMPILEDIR" gatk DetermineGermlineContigPloidy ...</code>, users can specify
- * the theano compilation directory (which is set to <code>$HOME/.theano</code> by default).  See theano documentation
- * at <a href="https://theano-pymc.readthedocs.io/en/latest/library/config.html">
- *     https://theano-pymc.readthedocs.io/en/latest/library/config.html</a>.
+ * <code>PYTENSOR_FLAGS="base_compiledir=PATH/TO/BASE_COMPILEDIR" gatk DetermineGermlineContigPloidy ...</code>, users can specify
+ * the PyTensor compilation directory (which is set to <code>$HOME/.pytensor</code> by default).  See PyTensor documentation
+ * at <a href="https://pytensor.readthedocs.io/en/latest/library/config.html">
+ *     https://pytensor.readthedocs.io/en/latest/library/config.html</a>.
  * </p>
  *
  * <h3>Tool run modes</h3>

src/main/java/org/broadinstitute/hellbender/tools/copynumber/GermlineCNVCaller.java

@@ -92,12 +92,12 @@
  * <p>OpenMP and MKL parallelism can be controlled by setting the <code>OMP_NUM_THREADS</code> and <code>MKL_NUM_THREADS</code>
  * environment variables, respectively.</p>
  *
- * <p>Advanced users may wish to set the <code>THEANO_FLAGS</code> environment variable to override the GATK theano
+ * <p>Advanced users may wish to set the <code>PYTENSOR_FLAGS</code> environment variable to override the GATK PyTensor
  * configuration. For example, by running
- * <code>THEANO_FLAGS="base_compiledir=PATH/TO/BASE_COMPILEDIR" gatk GermlineCNVCaller ...</code>, users can specify
- * the theano compilation directory (which is set to <code>$HOME/.theano</code> by default).  See theano documentation
- * at <a href="https://theano-pymc.readthedocs.io/en/latest/library/config.html">
- *     https://theano-pymc.readthedocs.io/en/latest/library/config.html</a>.
+ * <code>PYTENSOR_FLAGS="base_compiledir=PATH/TO/BASE_COMPILEDIR" gatk DetermineGermlineContigPloidy ...</code>, users can specify
+ * the PyTensor compilation directory (which is set to <code>$HOME/.pytensor</code> by default).  See PyTensor documentation
+ * at <a href="https://pytensor.readthedocs.io/en/latest/library/config.html">
+ *     https://pytensor.readthedocs.io/en/latest/library/config.html</a>.
  * </p>
  *
  * <h3>Resource usage</h3>

src/main/java/org/broadinstitute/hellbender/tools/copynumber/PostprocessGermlineCNVCalls.java

@@ -89,12 +89,12 @@
  * the python environment is already set up. Otherwise, the environment must be created and activated as described in the
  * main GATK README.md file.</p>
  *
- * <p>Advanced users may wish to set the <code>THEANO_FLAGS</code> environment variable to override the GATK theano
+ * <p>Advanced users may wish to set the <code>PYTENSOR_FLAGS</code> environment variable to override the GATK PyTensor
  * configuration. For example, by running
- * <code>THEANO_FLAGS="base_compiledir=PATH/TO/BASE_COMPILEDIR" gatk PostprocessGermlineCNVCalls ...</code>, users can specify
- * the theano compilation directory (which is set to <code>$HOME/.theano</code> by default).  See theano documentation
- * at <a href="https://theano-pymc.readthedocs.io/en/latest/library/config.html">
- *     https://theano-pymc.readthedocs.io/en/latest/library/config.html</a>.
+ * <code>PYTENSOR_FLAGS="base_compiledir=PATH/TO/BASE_COMPILEDIR" gatk DetermineGermlineContigPloidy ...</code>, users can specify
+ * the PyTensor compilation directory (which is set to <code>$HOME/.pytensor</code> by default).  See PyTensor documentation
+ * at <a href="https://pytensor.readthedocs.io/en/latest/library/config.html">
+ *     https://pytensor.readthedocs.io/en/latest/library/config.html</a>.
  * </p>
  *
  * <h3>Required inputs:</h3>

src/main/python/org/broadinstitute/hellbender/gcnvkernel/__init__.py

@@ -26,6 +26,6 @@
 from .tasks.inference_task_base import ConvergenceError
 from .utils import cli_commons, math
 
-assert pymc_version == "5.10.0", "gcnvkernel currently only supports PyMC 5.9.0; version found: {0}; " \
+assert pymc_version == "5.10.1", "gcnvkernel currently only supports PyMC 5.10.1; version found: {0}; " \
                                  "please upgrade or downgrade the PyMC module in your python environment " \
                                  "accordingly.".format(pymc_version)

src/main/python/org/broadinstitute/hellbender/gcnvkernel/models/model_denoising_calling.py

@@ -12,6 +12,7 @@
 import pytensor
 import pytensor.sparse as pst
 import pytensor.tensor as pt
+from pytensor.tensor.shape import unbroadcast
 from pymc import Normal, Deterministic, Potential, Lognormal, Exponential, TruncatedNormal, NegativeBinomial, Mixture
 
 from . import commons
@@ -1236,10 +1237,9 @@ def _get_update_log_class_emission_tk_pytensor_func(self) -> pytensor.compile.Fu
                                                   self.calling_config.num_copy_number_classes),
                                                  dtype=types.floatX)
 
-        # TODO unbroadcast seems to have been removed?
         # this converts TensorType from row to matrix in the edge case when number of intervals is equal to 2
         # (to avoid type mismatch later on)
-        # log_class_emission_cum_sum_tk = pytensor.shape.unbroadcast(log_class_emission_cum_sum_tk, 0)
+        log_class_emission_cum_sum_tk = unbroadcast(log_class_emission_cum_sum_tk, 0)
 
         def inc_log_class_emission_tk_except_for_first_interval(pi_jkc, q_c_tc, cum_sum_tk):
             """Adds the contribution of a given sample to the log class emission (symbolically).