AttributeError: 'module' object has no attribute 'logsumexp'

[williamtjordan]

Hello,

I am attempting to install this pipeline using Singularity and run the test profile as follows,

nextflow run nf-core/deepvariant -profile test,singularity

and am receiving the following error:

Traceback (most recent call last): File "/lscratch/1876264.sapelo2/Bazel.runfiles_gbFRRq/runfiles/com_google_deepvariant/deepvariant/make_examples.py", line 1188, in <module> tf.app.run() File "/opt/conda/envs/nf-core-deepvariant-1.0/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 125, in run _sys.exit(main(argv)) File "/lscratch/1876264.sapelo2/Bazel.runfiles_gbFRRq/runfiles/com_google_deepvariant/deepvariant/make_examples.py", line 1178, in main make_examples_runner(options) File "/lscratch/1876264.sapelo2/Bazel.runfiles_gbFRRq/runfiles/com_google_deepvariant/deepvariant/make_examples.py", line 1090, in make_examples_runner candidates, examples, gvcfs = region_processor.process(region) File "/lscratch/1876264.sapelo2/Bazel.runfiles_gbFRRq/runfiles/com_google_deepvariant/deepvariant/make_examples.py", line 806, in process self._initialize() File "/lscratch/1876264.sapelo2/Bazel.runfiles_gbFRRq/runfiles/com_google_deepvariant/deepvariant/make_examples.py", line 747, in _initialize self.options.variant_caller_options) File "/lscratch/1876264.sapelo2/Bazel.runfiles_gbFRRq/runfiles/com_google_deepvariant/deepvariant/variant_caller.py", line 123, in __init__ for n_total in range(self.max_cache_coverage + 1)] File "/lscratch/1876264.sapelo2/Bazel.runfiles_gbFRRq/runfiles/com_google_deepvariant/deepvariant/variant_caller.py", line 201, in _calc_reference_confidence log10_probs = genomics_math.normalize_log10_probs([-1.0, -1.0, -1.0]) File "/lscratch/1876264.sapelo2/Bazel.runfiles_gbFRRq/runfiles/com_google_deepvariant/third_party/nucleus/util/genomics_math.py", line 261, in normalize_log10_probs lse = log10sumexp(log10_probs) File "/lscratch/1876264.sapelo2/Bazel.runfiles_gbFRRq/runfiles/com_google_deepvariant/third_party/nucleus/util/genomics_math.py", line 201, in log10sumexp return LOG_10_OF_E * scipy.special.logsumexp( AttributeError: 'module' object has no attribute 'logsumexp'

Any information you could provide to fix the problem would be much appreciated.

[PhilPalmer]

Hi @williamtjordan, sorry the pipeline isn't working as expected.

I'm not immediately sure what is causing this error. Are you able to upload the .nextflow.log file which should be in the directory where you ran your command

[williamtjordan]

Hi Phil,

I have attached the nextflow log file in the working directory I used to run the command.

Thank you.

nextflow.log

[PhilPalmer]

Hi @williamtjordan,

Sorry for taking a short while to getting round to testing this.

I tested the first command you sent on an AWS instance (with a fresh install of Singularity) & it worked as expected (see nextflow.log)

So I am still not completely sure what's causing the issue. Here are some ideas and questions to debug it:

The issue may be to with your Singularity install. 1) How did you Singularity?

Also, from looking at the log file you sent it looks like you've made a private fork of the repo. 2) Have you made any changes to the pipeline here?

There is also this error:

Authentication is required but no CredentialsProvider has been registered

So perhaps you need to provide your GitHub credentials (see here)

It's very strange to have received this error though because the first process (preprocess_bam) seems to have run successfully. Perhaps it fails to properly download the reference files from S3 which are only required for the second process. 3) Do you have access to the working directory /home/wtj04047/work/75/6d0c3d6b79b4a91dbfc2a3887c635d to see what files are there?