Update bedtools version to 2.27.1

.travis.yml

@@ -18,8 +18,7 @@ services:
   - xvfb
 
 before_install:
-  - wget http://ftp.debian.org/debian/pool/main/b/bedtools/bedtools_2.21.0-1_amd64.deb
-  - sudo dpkg -i bedtools_2.21.0-1_amd64.deb
+  - sudo apt-get -y install bedtools=2.27.1+dfsg-4
   - sudo apt-key adv --keyserver keyserver.ubuntu.com --recv-keys E298A3A825C0D65DFD57CBB651716619E084DAB9
   - sudo add-apt-repository 'deb https://cloud.r-project.org/bin/linux/ubuntu bionic-cran35/'
   - sudo apt-get update

Dockerfile

@@ -35,9 +35,11 @@ RUN apt-get update \
     && rm -rf /var/lib/apt/lists/*
 
 # Install specific bedtools version
-RUN wget http://ftp.debian.org/debian/pool/main/b/bedtools/bedtools_2.21.0-1_amd64.deb \
-    && dpkg -i bedtools_2.21.0-1_amd64.deb \
-    && rm bedtools_2.21.0-1_amd64.deb
+RUN apt-get update \
+    && apt-get install -t unstable -y --no-install-recommends \
+        bedtools=2.27.1+dfsg-4 \
+    && rm -rf /tmp/downloaded_packages/ /tmp/*.rds \
+    && rm -rf /var/lib/apt/lists/*
 
 RUN apt-get update \
     && apt-get install -t unstable -y --no-install-recommends \

docs/source/intro.rst

@@ -44,7 +44,7 @@ Install non-Python dependencies:
 
 .. code-block:: bash
 
-   conda install -y bedtools==2.20.1
+   conda install -y bedtools==2.27.1
    conda install -y ucsc-twobittofa
    conda install -y bioconductor-deseq2
    conda install -y bioconductor-cqn
@@ -62,18 +62,15 @@ And then install the ``ngs-toolkit`` library with pip (available only through Py
 
 ``ngs_toolkit`` makes use of some non-Python dependencies.
 
+ - `bedtools <https://bedtools.readthedocs.io/en/latest/>`_: version should be at least 2.27.1
+
 The following are highly recommended only for some data or analysis types:
 
- - `bedtools <https://bedtools.readthedocs.io/en/latest/>`_: required for some ATAC/ChIP-seq functions. It is underlying the Python interface library to bedtools (pybedtools) which can be installed without bedtools.
  - `R <https://www.r-project.org/>`_ and some bioconductor libraries (optional):
-
    - `cqn <https://bioconductor.org/packages/release/bioc/html/cqn.html>`_ (optional): used for GC-content aware normalization of NGS data.
    - `DESeq2 <https://bioconductor.org/packages/release/bioc/html/DESeq2.html>`_ (optional): used for differential testing of genes/regulatory elements.
  - `Kent tools <https://github.com/ENCODE-DCC/kentUtils>`_ (optional): the 'twoBitToFa' binary from UCSC's Kent bioinformatics toolkit is used to convert between the 2bit and FASTA formats.
 
-.. note::
-   ``bedtools`` version should be below 2.24.0 (2.20.1 is used for testing)
-
 
 Testing
 =============================

ngs_toolkit/__init__.py

@@ -197,9 +197,26 @@ def setup_timestamping():
             exclude_functions=["from_dataframe"])
 
 
+def check_bedtools_version():
+    import pybedtools
+
+    # not existing
+    v = ".".join([str(x) for x in pybedtools.helpers.settings.bedtools_version])
+    msg = "Bedtools does not seem to be installed or is not in $PATH."
+    if v == '':
+        raise Exception(msg)
+
+    # too low version
+    msg = "Bedtools version '{}' is smaller than 2.27.".format(v)
+    msg = " Please upgrade to newer version."
+    if not pybedtools.helpers.settings._v_2_27_plus:
+        raise Exception(msg)
+
+
 # setup
 _LOGGER = setup_logger()
 _CONFIG = setup_config()
+check_bedtools_version()
 setup_graphic_preferences()
 setup_timestamping()
 

ngs_toolkit/atacseq.py

@@ -731,14 +731,14 @@ def measure_coverage(
                     output_dir, s.name + ".{}_coverage.sh".format(peak_set_name)
                 )
                 cmd = (
-                    "date\nbedtools coverage -counts -abam {bam} -b {bed} > {out}\ndate"
+                    "date\nbedtools coverage -counts -a {bed} -b {bam} > {out}\ndate"
                     .format(bam=s.aligned_filtered_bam, bed=sites.fn, out=output_file))
 
                 submit_job(
                     cmd, job_file,
                     jobname=job_name,
                     logfile=log_file,
-                    cores=1, mem=8000, time="04:00:00")
+                    cores=1, mem=8000, time="04:00:00", **kwargs)
 
     def collect_coverage(
         self,