autogenes segfault with Docker image 1.4

[thegrebe]

Hi there,

I am seeing a segfault when trying to use autogenes with Docker image 1.4:

ERROR ~ Error executing process > 'DECONVOLUTE (1)'

Caused by:
  Process `DECONVOLUTE (1)` terminated with an error exit status (139)


Command executed:

  runDeconvolutionNF.R 'sc1' '/home/sciliciumtheo/omnideconv_bench/data/single_cell/' 'bulk1' '/home/sciliciumtheo/omnideconv_bench/data/bulks/' 'autogenes' '/home/sciliciumtheo/omnideconv_bench/test_results_1.4_autogenes' 'false' '0' '0' 'hs' '4' '/home/sciliciumtheo/omnideconv_bench/pipeline'

Command exit status:
  139

Command output:
  [1] "Starting deconvolution script ..."
  [1] "Method: autogenes; sc-norm: cpm; bulk-norm: tpm"

Command error:
  [1] "Starting deconvolution script ..."
  Loading required package: snowfall
  Loading required package: snow
  Skipping python installation checks ...
  [1] "Method: autogenes; sc-norm: cpm; bulk-norm: tpm"
  Warning message:
  In dir.create(res_path, recursive = TRUE, showWarnings = TRUE) :
    '/home/sciliciumtheo/omnideconv_bench/test_results_1.4_autogenes/autogenes_sc1_cpm_bulk1_tpm_ct0_rep0' already exists
  fixed was chosen because multiple values were supplied for "mode"
  nusvr was chosen because multiple values were supplied for "model"
  Note: using the 'X' assay as the X matrix
  /root/.local/share/r-miniconda/envs/r-omnideconv/lib/python3.8/site-packages/anndata/_core/anndata.py:117: ImplicitModificationWarning: Transforming to str index.
    warnings.warn("Transforming to str index.", ImplicitModificationWarning)
  /root/.local/share/r-miniconda/envs/r-omnideconv/lib/python3.8/site-packages/anndata/_core/anndata.py:117: ImplicitModificationWarning: Transforming to str index.
    warnings.warn("Transforming to str index.", ImplicitModificationWarning)

   *** caught segfault ***
  address 0x7f2688290000, cause 'memory not mapped'

  Traceback:
   1: py_call_impl(callable, call_args$unnamed, call_args$named)
   2: (function (ngen = 2L, mode = "standard", nfeatures = NULL, weights = NULL,     objectives = NULL, seed = 0L, verbose = TRUE, ...) {    cl <- sys.call()    cl[[1L]] <- list2    call_args <- split_named_unnamed(eval(cl, parent.frame()))    result <- py_call_impl(callable, call_args$unnamed, call_args$named)    if (py_get_convert(callable))         result <- py_to_r(result)    if (is.null(result))         invisible(result)    else result})(ngen = 5000L, weights = list(-1, 1), objectives = list("correlation",     "distance"), verbose = TRUE, seed = 0L, mode = "fixed", population_size = 100L,     offspring_size = 100L, crossover_pb = 0.7, mutation_pb = 0.3,     nfeatures = 500L, mutate_flip_pb = 0.001)
   3: do.call(ag$optimize, params)
   4: omnideconv::deconvolute_autogenes(single_cell_object = sc_matrix,     cell_type_annotations = annotations, bulk_gene_expression = bulk_matrix,     max_iter = 1e+06, ngen = 5000, verbose = TRUE)
   5: deconvolution_workflow_general(sc_matrix, sc_celltype_annotations,     "normal", sc_dataset, sc_norm, sc_batch, signature, method,     bulk_matrix, bulk_name, bulk_norm, ncores, res_path, rmbatch_S_mode = s_mode,     rmbatch_B_mode = b_mode, baseDir = baseDir)
   6: system.time({    deconvolution <- deconvolution_workflow_general(sc_matrix,         sc_celltype_annotations, "normal", sc_dataset, sc_norm,         sc_batch, signature, method, bulk_matrix, bulk_name,         bulk_norm, ncores, res_path, rmbatch_S_mode = s_mode,         rmbatch_B_mode = b_mode, baseDir = baseDir)})
  An irrecoverable exception occurred. R is aborting now ...
  .command.sh: line 2:     8 Segmentation fault      (core dumped) runDeconvolutionNF.R 'sc1' '/home/sciliciumtheo/omnideconv_bench/data/single_cell/' 'bulk1' '/home/sciliciumtheo/omnideconv_bench/data/bulks/' 'autogenes' '/home/sciliciumtheo/omnideconv_bench/test_results_1.4_autogenes' 'false' '0' '0' 'hs' '4' '/home/sciliciumtheo/omnideconv_bench/pipeline'

Work dir:
  /home/sciliciumtheo/omnideconv_bench/work/2d/45a31dffb06873300d15cad7165360

Container:
  alexd13/omnideconv_benchmark:1.4

Tip: view the complete command output by changing to the process work dir and entering the command `cat .command.out`

 -- Check '.nextflow.log' file for details

autogenes ran normally with Docker image with latest tag.

Best,
Theo

[alex-d13]

This issue is also (hopefully) being discussed here: rstudio/reticulate#1775