static_all 1.1.3 seer/kmds segfault at end

[tseemann]

It prints the help, then it pauses for 1 second, then it dumps core. The kmds command does the same thing. Downloaded from: https://github.com/johnlees/seer/releases/download/v1.1.3/seer_v1.1.3_static_all.tar.gz The same happens with the _static version too. Note is says 1.2alpha4, not 1.1.3 I'm on glibc-2.17, RHEL 7.2, kernel 3.10

seer --version
seer: sequence element enrichment analysis
1.2alpha4

<pause 1 second>

Segmentation fault (core dumped)

It returns error code 139

seer -h
seer: sequence element enrichment analysis

Required options:
  -k [ --kmers ] arg       dsm kmer output file
  -p [ --pheno ] arg       .pheno metadata

Covariate options:
  --struct arg             mds values from kmds
  --covar_file arg         file containing covariates
  --covar_list arg         list of columns covariates to use. Format is 1,2q,3
                           (use q for quantitative)

Performance options:
  --threads arg (=1)       number of threads. Suggested: 72

Filtering options:
  --no_filtering           turn off all filtering and perform tests on all
                           kmers input
  --max_length arg (=100)  maximum kmer length
  --maf arg (=0.01)        minimum kmer frequency
  --min_words arg          minimum kmer occurences. Overrides --maf
  --chisq arg (=10e-5)     p-value threshold for initial chi squared test. Set
                           to 1 to show all
  --pval arg (=10e-8)      p-value threshold for final logistic test. Set to 1
                           to show all

Other options:
  --print_samples          print lists of samples significant kmers were found
                           in
  --version                prints version and exits
  -h [ --help ]            full help message

 #<pause>

Segmentation fault (core dumped)

[johnlees]

As far as I can tell the programs run correctly, and the segfault is only caused on exit. This only seems to happen with the statically compiled version.

I realise this isn't ideal, but I don't know what would be causing it, and all other functionality seems ok. When I get a moment, I'm planning to sort out these various compiler/installation issues by integrating seer into bioconda (Issue #48).

[tseemann]

I had commenced a Homebrew package but there was a few Makefile issues that made it challenging. Moving to pkg-config would help, or autotools (but painful). It's still a goal to get it to compile from source.

I suspect if you manage to pass Conda's strict tests i'll be able to put it in Brew.

Thanks for looking into it.

[johnlees]

We've had a go at this (with the help of @Slugger70), but need a newer compiler for it to work: https://github.com/bioconda/bioconda-recipes/pull/5005

This is being added to bioconda at the moment, so I'm hopeful that we can have seer in bioconda (and docker) within a couple of weeks!

[tseemann]

@johnlees yeah, @slugger70 reminded me of the old build system that Conda uses which doesn't have C++11 or C++14 support yet :(