Updating install reqs for #886

docker/Dockerfile

@@ -57,8 +57,7 @@ RUN apt-get -qq update \
   # Required for R
 	gfortran \
   # Required for rpy2
-  r-base-dev \
-  r-base \
+  r-cran-rjava \
   # Required for R-package devtools (which is required for SJD)
   libharfbuzz-dev \
   libfribidi-dev \

docker/install_bioc.R

@@ -1,5 +1,6 @@
 #!/usr/bin/env Rscript --vanilla
 
 install.packages(c("BiocManager", "devtools"), dependencies=TRUE, repos="http://lib.stat.cmu.edu/R/CRAN/")
+BiocManager::install(version = "3.19")  # required for R 4.4.0
 BiocManager::install(c("genesofeve/projectR", "biomaRt"), ask=FALSE)
 library(devtools); install_github("CHuanSite/SJD")

docker/install_bioc.sh

@@ -5,7 +5,7 @@ Rver="${Rmaj}.4.0"
 
 current_dir=$(pwd)
 
-curl -s -L http://lib.stat.cmu.edu/R/CRAN/src/base/${Rmaj}/${Rver}.tar.gz | tar xzv -C /opt
+curl -s -L http://lib.stat.cmu.edu/R/CRAN/src/base/${Rmaj}/${Rver}.tar.gz | tar xzv -C /opt || exit 1
 cd /opt/${Rver}
 /opt/${Rver}/configure --with-readline=no --enable-R-shlib --enable-BLAS-shlib --with-x=no || exit 1
 make || exit 1

docker/requirements.txt

@@ -20,7 +20,7 @@ more_itertools==9.0.0
 mysql-connector-python==8.0.20
 numba==0.58.1
 numexpr==2.8.4
-numpy==1.26.0
+numpy==1.26.4
 opencv-python==4.5.5.64
 openpyxl==3.1.5
 pandas==2.2.1
@@ -29,7 +29,7 @@ pika==1.3.1
 plotly==5.6.0
 python-dotenv==0.20.0
 requests==2.31.0
-rpy2==3.5.1 # 3.5.2 and up gives errors with rpy2py and py2rpy
+rpy2==3.5.16
 sanic
 scanpy==1.10.1
 scikit-learn==1.0.2

docs/setup.python.md

@@ -60,7 +60,7 @@ Check the requirement.txt file in <git_repo_root>/docker for the latest packages
       mysql-connector-python==8.0.20 \
       numba==0.58.1 \
       numexpr==2.8.4 \
-      numpy==1.26.0 \
+      numpy==1.26.4 \
       opencv-python==4.5.5.64 \
       openpyxl==3.1.5 \
       pandas==2.2.1 \

services/projectr/Dockerfile

@@ -11,8 +11,7 @@ RUN apt-get -qq update \
     # Required for R
 	gfortran \
     # Required for rpy2
-    r-base-dev \
-    r-base \
+    r-cran-rjava \
     # Required for R-package devtools (which is required for SJD)
     libharfbuzz-dev \
     libfribidi-dev \

services/projectr/install_bioc.R

@@ -1,5 +1,6 @@
 #!/usr/bin/env Rscript --vanilla
 
 install.packages(c("BiocManager", "devtools"), dependencies=TRUE, repos="http://lib.stat.cmu.edu/R/CRAN/")
+BiocManager::install(version = "3.19")  # required for R 4.4.0
 BiocManager::install(c("genesofeve/projectR", "biomaRt"), ask=FALSE)
 library(devtools); install_github("CHuanSite/SJD")

services/projectr/install_bioc.sh

@@ -5,7 +5,7 @@ Rver="${Rmaj}.4.0"
 
 current_dir=$(pwd)
 
-curl -s -L http://lib.stat.cmu.edu/R/CRAN/src/base/${Rmaj}/${Rver}.tar.gz | tar xzv -C /opt
+curl -s -L http://lib.stat.cmu.edu/R/CRAN/src/base/${Rmaj}/${Rver}.tar.gz | tar xzv -C /opt || exit 1
 cd /opt/${Rver}
 /opt/${Rver}/configure --with-readline=no --enable-R-shlib --enable-BLAS-shlib --with-x=no || exit 1
 make || exit 1

services/projectr/requirements.txt

@@ -1,8 +1,6 @@
 Flask==3.0.0
 gunicorn==20.1.0
-rpy2==3.5.1 # 3.5.2 and up gives errors with rpy2py and py2rpy
-#rpy2==3.5.16
-#pandas==2.2.1
+rpy2==3.5.16
+pandas==2.2.1
 numpy==1.26.4   # https://stackoverflow.com/a/78641304
-pandas==1.4.1
 google-cloud-logging

services/projectr/rfuncs.py

@@ -16,6 +16,10 @@
 from rpy2.robjects.conversion import localconverter
 from rpy2.robjects.vectors import StrVector
 
+# If running locally, need to ensure that multiple concurrent R calls do not conflict
+from rpy2.rinterface_lib import openrlib
+
+
 class RError(Exception):
     """Error based on issues that would manifest in any particular R-language call."""
     def __init__(self, message="") -> None:
@@ -45,51 +49,54 @@ def run_projectR_cmd(target_df, loading_df, algorithm):
     Return Pandas dataframe of the projectR output
     """
 
-    # Convert from pandas dataframe to R data.frame
-    with localconverter(ro.default_converter + pandas2ri.converter):
-        target_r_df = ro.conversion.py2rpy(target_df)
-        loading_r_df = ro.conversion.py2rpy(loading_df)
-
-    # data.frame to matrix (projectR has no data.frame signature)
-    target_r_matrix = convert_r_df_to_r_matrix(target_r_df)
-    loading_r_matrix = convert_r_df_to_r_matrix(loading_r_df)
-
-    # Assign Rownames to each matrix
-    # I don't know why but using ro.StrVector makes rpy2py fail where the output df is an incompatible class
-    # Guessing that there are some non-strings mixed into the indexes
-    target_r_matrix.rownames = StrVector(target_df.index)
-    loading_r_matrix.rownames = StrVector(loading_df.index)
-
-    # The NMF projectR method signature is based on the LinearEmbeddedMatrix class,
-    # Which has a featureLoadings property. That matrix is loaded and the default
-    # projectR signature is returned and used. So we can just pass the matrix as-is.
-    # https://rdrr.io/bioc/SingleCellExperiment/man/LinearEmbeddingMatrix.html
-
-    # Run project R command.  Get projectionPatterns matrix
-    try:
-        if algorithm == "nmf":
-            projectR = importr('projectR')
-            projection_patterns_r_matrix = projectR.projectR(data=target_r_matrix, loadings=loading_r_matrix, full=False)
-        elif algorithm == "fixednmf":
-            sjd = importr('SJD')
-            loading_list = ro.ListVector({"genesig": loading_r_matrix})
-
-            projection = sjd.projectNMF(proj_dataset=target_r_matrix, proj_group=True, list_component=loading_list)
-            projection_patterns_r_matrix = projection.rx2("proj_score_list").rx2("genesig")
-        else:
-            raise ValueError("Algorithm {} is not supported".format(algorithm))
-    except Exception as e:
-        # print stacktrace with line numbers
-        traceback.print_exc(file=sys.stderr)
-        raise RError("Error: Could not run projectR command.\tReason: {}".format(str(e)))
-
-    # matrix back to data.frame
-    projection_patterns_r_df = convert_r_matrix_to_r_df(projection_patterns_r_matrix)
-
-    # Convert from R data.frame to pandas dataframe
-    with localconverter(ro.default_converter + pandas2ri.converter):
-        projection_patterns_df = ro.conversion.rpy2py(projection_patterns_r_df)
-
-    return projection_patterns_df
+    # Ensure multithreading if running locally -> https://rpy2.github.io/doc/v3.5.x/html/rinterface.html#multithreading
+    with openrlib.rlock:
+
+        # Convert from pandas dataframe to R data.frame
+        with localconverter(ro.default_converter + pandas2ri.converter):
+            target_r_df = ro.conversion.py2rpy(target_df)
+            loading_r_df = ro.conversion.py2rpy(loading_df)
+
+        # data.frame to matrix (projectR has no data.frame signature)
+        target_r_matrix = convert_r_df_to_r_matrix(target_r_df)
+        loading_r_matrix = convert_r_df_to_r_matrix(loading_r_df)
+
+        # Assign Rownames to each matrix
+        # I don't know why but using ro.StrVector makes rpy2py fail where the output df is an incompatible class
+        # Guessing that there are some non-strings mixed into the indexes
+        target_r_matrix.rownames = StrVector(target_df.index)
+        loading_r_matrix.rownames = StrVector(loading_df.index)
+
+        # The NMF projectR method signature is based on the LinearEmbeddedMatrix class,
+        # Which has a featureLoadings property. That matrix is loaded and the default
+        # projectR signature is returned and used. So we can just pass the matrix as-is.
+        # https://rdrr.io/bioc/SingleCellExperiment/man/LinearEmbeddingMatrix.html
+
+        # Run project R command.  Get projectionPatterns matrix
+        try:
+            if algorithm == "nmf":
+                projectR = importr('projectR')
+                projection_patterns_r_matrix = projectR.projectR(data=target_r_matrix, loadings=loading_r_matrix, full=False)
+            elif algorithm == "fixednmf":
+                sjd = importr('SJD')
+                loading_list = ro.ListVector({"genesig": loading_r_matrix})
+
+                projection = sjd.projectNMF(proj_dataset=target_r_matrix, proj_group=True, list_component=loading_list)
+                projection_patterns_r_matrix = projection.rx2("proj_score_list").rx2("genesig")
+            else:
+                raise ValueError("Algorithm {} is not supported".format(algorithm))
+        except Exception as e:
+            # print stacktrace with line numbers
+            traceback.print_exc(file=sys.stderr)
+            raise RError("Error: Could not run projectR command.\tReason: {}".format(str(e)))
+
+        # matrix back to data.frame
+        projection_patterns_r_df = convert_r_matrix_to_r_df(projection_patterns_r_matrix)
+
+        # Convert from R data.frame to pandas dataframe
+        with localconverter(ro.default_converter + pandas2ri.converter):
+            projection_patterns_df = ro.conversion.rpy2py(projection_patterns_r_df)
+
+        return projection_patterns_df