diff --git a/.dockerignore b/.dockerignore
new file mode 100644
index 0000000..db9a224
--- /dev/null
+++ b/.dockerignore
@@ -0,0 +1,5 @@
+/examples/
+/cwls/
+/git-hooks/
+/*.*
+/LICENSE
diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..47def24
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1 @@
+/env/
diff --git a/.travis.yml b/.travis.yml
index 0f265bb..272317b 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -8,7 +8,25 @@ services:
   - docker
 
 install:
- - docker build -t dockstore-cgpmap .
+ - virtualenv -p python3 venv
+ - source venv/bin/activate
+ - pip install html5lib cwltool
 
 script:
+ - set -e
+ - echo 'Validate CWL file(s)'
+ - cwltool --validate Dockstore.cwl
+ - cwltool --validate cwls/cgpmap-bamBaiOut.cwl
+ - cwltool --validate cwls/cgpmap-bamCsiOut.cwl
+ - cwltool --validate cwls/cgpmap-cramOut.cwl
+ - echo 'Build and check docker image'
+ - docker build -t dockstore-cgpmap .
  - docker images | grep -c dockstore-cgpmap
+ - echo 'Verify program from each inherited package is found (dockstore-cgpbigwig)'
+ - docker run -ti --rm dockstore-cgpmap bwjoin --version
+ - echo 'Verify a program from each new package is found (dockstore-cgpmap)'
+ - docker run -ti --rm dockstore-cgpmap ds-cgpmap.pl -h
+ - docker run -ti --rm dockstore-cgpmap bwa_mem.pl -version
+ - docker run -ti --rm dockstore-cgpmap bammarkduplicates2 --version
+ - docker run -ti --rm dockstore-cgpmap samtools --version
+ - docker run -ti --rm dockstore-cgpmap bash -c 'bwa 2>&1 | grep Version'
diff --git a/CHANGES.md b/CHANGES.md
index d5ea2b3..25005e5 100644
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -1,35 +1,68 @@
-### 2.0.1
+# CHANGES
+
+## 3.0.0
+
+* Drop pre/post exec functions in `mapping.sh`
+* Add ability to pair a groups file with fastq inputs to add info to readgroups
+in final BAM/CRAM files (PCAP-core).
+* BWA 0.7.17 - bug fixes for future alpine building.
+* Biobambam2 2.0.84 - via pre-compiled versions
+* HTSlib + Samtools 1.7
+* PCAP-core 4.1.1
+  * mismatchQc options added
+* cgpBigWig 1.0.0 (via dockstore-cgpbigwig 2.0.0)
+* Examples moved to more useful naming, now have:
+  * `examples/bamOutput` and `examples/cramOutput` each containing:
+    * `bam_input.json`
+    * `fastq_gz_input.json` - with yaml groupinfo file example.
+* Multiple CWL descriptors one for BAM output, another for CRAM, legacy version retained.
+  * Ensure you use the correct `json` examples.
+* Now using secondaryFiles for outputs to reduce repetitive entries in cwl and json.
+  * Bugfix to dockstore tool needed for output provisioning, 1.3.6+
+
+## 2.0.1
+
 * Test data in `examples/sample_configs.local.json` moved to a non-expiring location.
 
-### 2.0.0
+## 2.0.0
+
 * PCAP-core forked to cancerit and all legacy PCAWG code removed.
 * Update to cgpBigWig/libBigWig to handle bug detected in ASCAT.
 * First layer of streamlined install process to reduce build time of dependant images.
 * Biobambam2 now building from source (previously picked up precompiled 'etch').
 
-### 1.0.8
+## 1.0.8
+
 Fix in PCAP-core to handle passing of sample name when input BAM has no SM tag in header.
 
-### 1.0.7
+## 1.0.7
+
 HTSlib upgades in toolset for consistency.
 
-### 1.0.6
+## 1.0.6
+
 Bump PCAP-core version to fix fastq input handling
 
-### 1.0.5
+## 1.0.5
+
 Adds travis-ci
 
-### 1.0.4
+## 1.0.4
+
 Bad versions in 1.0.3
 
-### 1.0.3
+## 1.0.3
+
 Base PCAP-core upgraded to improve mapping through-put and CPU use.
 
-### 1.0.2
+## 1.0.2
+
 Added build badges
 
-### 1.0.1
+## 1.0.1
+
 Fix CWL to get description to display on Dockstore.org
 
-### 1.0.0
+## 1.0.0
+
 Initial release.  Fully functional.
diff --git a/Dockerfile b/Dockerfile
index 596bba7..94fd412 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,36 +1,86 @@
-FROM  ubuntu:14.04
+FROM  quay.io/wtsicgp/dockstore-cgpbigwig:2.0.0 as builder
 
-MAINTAINER  keiranmraine@gmail.com
+USER  root
 
-LABEL uk.ac.sanger.cgp="Cancer Genome Project, Wellcome Trust Sanger Institute" \
-      version="2.0.3" \
-      description="The CGP mapping pipeline for dockstore.org"
+RUN apt-get -yq update
+RUN apt-get install -yq --no-install-recommends\
+  apt-transport-https\
+  locales\
+  curl\
+  ca-certificates\
+  libperlio-gzip-perl\
+  make\
+  bzip2\
+  gcc\
+  psmisc\
+  time\
+  zlib1g-dev\
+  libbz2-dev\
+  liblzma-dev\
+  libcurl4-gnutls-dev\
+  libncurses5-dev\
+  nettle-dev\
+  libp11-kit-dev\
+  libtasn1-dev\
+  libgnutls-dev
 
-USER  root
+RUN locale-gen en_US.UTF-8
+RUN update-locale LANG=en_US.UTF-8
 
 ENV OPT /opt/wtsi-cgp
-ENV PATH $OPT/bin:$PATH
+ENV PATH $OPT/bin:$OPT/biobambam2/bin:$PATH
 ENV PERL5LIB $OPT/lib/perl5
 ENV LD_LIBRARY_PATH $OPT/lib
-
-## USER CONFIGURATION
-RUN adduser --disabled-password --gecos '' ubuntu && chsh -s /bin/bash && mkdir -p /home/ubuntu
+ENV LC_ALL en_US.UTF-8
+ENV LANG en_US.UTF-8
 
 RUN mkdir -p $OPT/bin
 
-ADD scripts/mapping.sh $OPT/bin/mapping.sh
-ADD scripts/ds-wrapper.pl $OPT/bin/ds-wrapper.pl
-RUN chmod a+x $OPT/bin/mapping.sh $OPT/bin/ds-wrapper.pl
+ADD build/opt-build.sh build/
+RUN bash build/opt-build.sh $OPT
 
-ADD build/apt-build.sh build/
-RUN bash build/apt-build.sh
+FROM  ubuntu:16.04
 
-ADD build/perllib-build.sh build/
-RUN bash build/perllib-build.sh
+MAINTAINER  cgphelp@sanger.ac.uk
 
-ADD build/opt-build.sh build/
-ADD build/biobambam2-build.sh build/
-RUN bash build/opt-build.sh $OPT
+LABEL vendor="Cancer, Ageing and Somatic Mutation, Wellcome Trust Sanger Institute"
+LABEL uk.ac.sanger.cgp.description="PCAP-core for dockstore.org"
+LABEL uk.ac.sanger.cgp.version="3.0.0"
+
+RUN apt-get -yq update
+RUN apt-get install -yq --no-install-recommends\
+  apt-transport-https\
+  locales\
+  curl\
+  ca-certificates\
+  libperlio-gzip-perl\
+  bzip2\
+  psmisc\
+  time\
+  zlib1g\
+  liblzma5\
+  libncurses5\
+  p11-kit
+
+RUN locale-gen en_US.UTF-8
+RUN update-locale LANG=en_US.UTF-8
+
+ENV OPT /opt/wtsi-cgp
+ENV PATH $OPT/bin:$OPT/biobambam2/bin:$PATH
+ENV PERL5LIB $OPT/lib/perl5
+ENV LD_LIBRARY_PATH $OPT/lib
+ENV LC_ALL en_US.UTF-8
+ENV LANG en_US.UTF-8
+
+RUN mkdir -p $OPT
+COPY --from=builder $OPT $OPT
+
+ADD scripts/mapping.sh $OPT/bin/mapping.sh
+ADD scripts/ds-cgpmap.pl $OPT/bin/ds-cgpmap.pl
+RUN chmod a+x $OPT/bin/mapping.sh $OPT/bin/ds-cgpmap.pl
+
+## USER CONFIGURATION
+RUN adduser --disabled-password --gecos '' ubuntu && chsh -s /bin/bash && mkdir -p /home/ubuntu
 
 USER    ubuntu
 WORKDIR /home/ubuntu
diff --git a/Dockstore.cwl b/Dockstore.cwl
index 82335e3..0b33170 100644
--- a/Dockstore.cwl
+++ b/Dockstore.cwl
@@ -9,23 +9,19 @@ label: "CGP BWA-mem mapping flow"
 cwlVersion: v1.0
 
 doc: |
-    ![build_status](https://quay.io/repository/wtsicgp/dockstore-cgpmap/status)
-    A Docker container for the CGP BWA-mem mapping flow. See the [dockstore-cgpmap](https://github.com/cancerit/dockstore-cgpmap) website for more information.
+  Please use one of the new tools for v3+:
 
-dct:creator:
-  "@id": "http://orcid.org/0000-0002-5634-1539"
-  foaf:name: Keiran M Raine
-  foaf:mbox: "keiranmraine@gmail.com"
+    * [dockstore-cgpmap/cgpmap-bamOut](https://dockstore.org/containers/quay.io%2Fwtsicgp%2Fdockstore-cgpmap%2Fcgpmap-bamOut)
+    * [dockstore-cgpmap/cgpmap-cramOut](https://dockstore.org/containers/quay.io%2Fwtsicgp%2Fdockstore-cgpmap%2Fcgpmap-cramOut)
+
+  ![build_status](https://quay.io/repository/wtsicgp/dockstore-cgpmap/status)
+  A Docker container for PCAP-core. See the [dockstore-cgpmap](https://github.com/cancerit/dockstore-cgpmap) website for more information.
 
 requirements:
-  - class: DockerRequirement
-    dockerPull: "quay.io/wtsicgp/dockstore-cgpmap:2.0.3"
+  - $mixin: cwls/mixins/requirements.yml
 
 hints:
-  - class: ResourceRequirement
-    coresMin: 1 # works but long, 8 recommended
-    ramMin: 15000 # good for WGS human ~30-60x
-    outdirMin: 5000000 # unlikely any BAM processing would be possible in less
+  - $mixin: cwls/mixins/hints.yml
 
 inputs:
   reference:
@@ -52,33 +48,40 @@ inputs:
       position: 3
       separate: true
 
-  scramble:
-    type: string?
-    doc: "Options to pass to scramble when generating CRAM output, see scramble docs"
-    default: ''
-    inputBinding:
-      prefix: -scramble
-      position: 4
-      separate: true
-      shellQuote: true
-
   bwa:
     type: string?
     default: ' -Y -K 100000000'
     doc: "Mapping and output parameters to pass to BWA-mem, see BWA docs, default ' -Y -K 100000000'"
     inputBinding:
       prefix: -bwa
-      position: 5
+      position: 4
       separate: true
       shellQuote: false
 
-  cram:
+  groupinfo:
+    type: File?
+    doc: "Readgroup metadata file for FASTQ inputs"
+    inputBinding:
+      prefix: -groupinfo
+      position: 5
+      separate: true
+
+  mmqc:
     type: boolean
-    doc: "Set if output should be in CRAM format instead of BAM, see 'scramble' for tuning parameters."
+    doc: "Apply mismatch QC to reads following duplicate marking."
     inputBinding:
-      prefix: -cram
+      prefix: -qc
       position: 6
 
+  mmqcfrac:
+    type: float?
+    default: 0.05
+    doc: "Mismatch fraction to set as max before failing a read [0.05]"
+    inputBinding:
+      prefix: -qcf
+      position: 7
+      separate: true
+
   bams_in:
     type:
     - 'null'
@@ -86,7 +89,7 @@ inputs:
       items: File
     doc: "Can be BAM, CRAM, fastq (paired or interleaved), BAM/CRAM can be mixed together but not FASTQ."
     inputBinding:
-      position: 7
+      position: 8
 
 outputs:
   out_bam:
@@ -119,4 +122,19 @@ outputs:
     outputBinding:
       glob: $(inputs.sample).bam.maptime
 
-baseCommand: ["/opt/wtsi-cgp/bin/ds-wrapper.pl"]
+baseCommand: ["/opt/wtsi-cgp/bin/ds-cgpmap.pl"]
+
+$schemas:
+  - http://schema.org/docs/schema_org_rdfa.html
+
+$namespaces:
+  s: http://schema.org/
+
+s:codeRepository: https://github.com/cancerit/dockstore-cgpmap
+s:license: https://spdx.org/licenses/AGPL-3.0-only
+
+s:author:
+  - class: s:Person
+    s:identifier: https://orcid.org/0000-0002-5634-1539
+    s:email: mailto:cgphelp@sanger.ac.uk
+    s:name: Keiran Raine
diff --git a/README.md b/README.md
index 1412fe8..1b81d02 100644
--- a/README.md
+++ b/README.md
@@ -1,31 +1,109 @@
-dockstore-cgpmap
-======
-`dockstore-cgpmap` provides a complete multi threaded BWA mem mapping workflow.  This has been packaged specifically for use with the [Dockstore.org](https://dockstore.org/) framework.
+# dockstore-cgpmap
 
-[![Join the chat at https://gitter.im/dockstore-cgpmap/general](https://badges.gitter.im/dockstore-cgpmap/general.svg)](https://gitter.im/dockstore-cgpmap/general?utm_source=badge&utm_medium=badge&utm_campaign=pr-badge&utm_content=badge)
+`dockstore-cgpmap` provides a complete multi threaded BWA mem mapping workflow.  This has been
+packaged specifically for use with the [Dockstore.org](https://dockstore.org/) framework.
 
-[![Docker Repository on Quay](https://quay.io/repository/wtsicgp/dockstore-cgpmap/status "Docker Repository on Quay")](https://quay.io/repository/wtsicgp/dockstore-cgpmap)
+[![Gitter Badge][gitter-svg]][gitter-badge]
 
-[![Build Status](https://travis-ci.org/cancerit/dockstore-cgpmap.svg?branch=master)](https://travis-ci.org/cancerit/dockstore-cgpmap) : master  
-[![Build Status](https://travis-ci.org/cancerit/dockstore-cgpmap.svg?branch=develop)](https://travis-ci.org/cancerit/dockstore-cgpmap) : develop
+[![Quay Badge][quay-status]][quay-repo]
+
+| Master                                        | Develop                                         |
+| --------------------------------------------- | ----------------------------------------------- |
+| [![Master Badge][travis-master]][travis-base] | [![Develop Badge][travis-develop]][travis-base] |
+
+<!-- TOC depthFrom:2 depthTo:6 withLinks:1 updateOnSave:1 orderedList:0 -->
+
+- [Supported input formats:](#supported-input-formats)
+- [Options for customisation:](#options-for-customisation)
+- [Usable Cores](#usable-cores)
+- [Other uses](#other-uses)
+	- [Native docker](#native-docker)
+	- [Singularity](#singularity)
+- [Test data](#test-data)
+- [Release process](#release-process)
+- [LICENCE](#licence)
+
+<!-- /TOC -->
+
+## Supported input formats:
 
-## Supports input in following formats:
 * Multiple BAM
 * Multiple CRAM
 * Multiple fastq[.gz] (paired or interleaved)
-  * Please see [PCAP-core/bin/bwa_mem.pl](https://github.com/cancerit/PCAP-core/blob/master/bin/bwa_mem.pl) for formatting of file names.
+  * Please see [PCAP-core/bin/bwa_mem.pl][bwa-mem.pl]
+for formatting of file names.
 
 ## Options for customisation:
 
 * BWA specific mapping parameters (defaults are based on attempts at a global standard).
 * Optionally output CRAM (scramble parameters can be modified)
 
-# Test data
-The `examples/sample_configs.local.json` contains test data that can be used to verify the tool.
+## Usable Cores
+
+When running outside of a docker container you can set the number of CPUs via:
+
+* `export CPU=N`
+* `-threads|-t` option of `ds-cgpmap.pl`
+
+If not set detects available cores on system.
+
+## Other uses
 
-You can find expected outputs on the Sanger Institute FTP site: [dockstore-cgpmap-expected.tar.gz](ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/expected/dockstore-cgpmap-expected.tar.gz)
+### Native docker
+
+All of the tools installed as part of [PCAP-core][pcap-core] are available for direct use.
+
+```
+export CGPMAP_VER=X.X.X
+docker pull quay.io/wtsicgp/dockstore-cgpmap:$CGPMAP_VER
+# interactive session
+docker --rm -ti [--volume ...] quay.io/wtsicgp/dockstore-cgpmap:$CGPMAP_VER bash
+```
+
+### Singularity
+
+The resulting docker container has been tested with Singularity.  The command to exec is:
+
+```
+ds-cgpmap.pl -h
+```
 
-This project includes the C program `diff_bams` that can be used to compare the generated BAM file to the one in the archive:
+Expected use would be along the lines of:
+
+```
+export CGPMAP_VER=X.X.X
+singularity pull docker://quay.io/wtsicgp/dockstore-cgpmap:$CGPMAP_VER
+
+singularity exec\
+ --workdir /.../workspace  \
+ --home /.../workspace:/home  \
+ --bind /.../ref/human:/var/spool/ref:ro  \
+ --bind /.../example_data/cgpmap/insilico_21:/var/spool/data:ro  \
+ dockstore-cgpmap-${CGPMAP_VER}.simg  \
+ ds-cgpmap.pl  \
+ -r /var/spool/ref/core_ref_GRCh37d5.tar.gz  \
+ -i /var/spool/ref/bwa_idx_GRCh37d5.tar.gz  \
+ -s SOMENAME  \
+ -t 6 \
+ /var/spool/data/\*.bam
+```
+
+For a system automatically attaching _all local mount points_ (not default singularity behaviour)
+you need not specify any `exec` params (workdir, home, bind) but you should specify the `-outdir`
+option for `ds-cgpmap.pl` to prevent data being written to your home directory.
+
+By default results are written to the home directory of the container so ensure you bind
+a large volume and set the `-home` variable.  As indicated above the location can be overridden
+via the options of `ds-cgpmap.pl`
+
+## Test data
+
+The `examples/` contains test data that can be used to verify the tool.
+
+You can find expected outputs on the Sanger Institute FTP site (bam output): [dockstore-cgpmap-expected.tar.gz][cgpmap-expected]
+
+This project includes the C program `diff_bams` that can be used to compare the generated BAM file
+to the one in the archive:
 
 ```bash
 $ export CGPMAP_TAG=0.2.0
@@ -45,24 +123,24 @@ Reference sequence order passed
 Matching records: 1000001
 ```
 
-Release process
-===============
+## Release process
+
 This project is maintained using HubFlow.
 
 1. Make appropriate changes
-2. Bump version in `Dockerfile` and `Dockstore.cwl`
-3. Push changes
-4. Check state on Travis
-5. Generate the release (add notes to GitHub)
-6. Confirm that image has been built on [quay.io](https://quay.io/repository/wtsicgp/dockstore-cgpmap?tab=builds)
-7. Update the [dockstore](https://dockstore.org/containers/quay.io/wtsicgp/dockstore-cgpmap) entry, see [their docs](https://dockstore.org/docs/getting-started-with-dockstore).
+1. Bump version in `Dockerfile` and `cwls/mixins/requirements.yml`
+1. Push changes
+1. Check state on Travis
+1. Generate the release (add notes to GitHub)
+1. Confirm that image has been built on [quay.io][quay-builds]
+1. Update the [dockstore][dockstore-cgpmap] entry, see [their docs][dockstore-get-started].
 
-LICENCE
-=======
+## LICENCE
 
-Copyright (c) 2016-2017 Genome Research Ltd.
+```
+Copyright (c) 2016-2018 Genome Research Ltd.
 
-Author: Cancer Genome Project <cgpit@sanger.ac.uk>
+Author: CASM/Cancer IT <cgphelp@sanger.ac.uk>
 
 This file is part of dockstore-cgpmap.
 
@@ -88,3 +166,27 @@ reads ‘Copyright (c) 2005, 2007, 2008, 2009, 2011, 2012’ and a copyright
 statement that reads ‘Copyright (c) 2005-2012’ should be interpreted as being
 identical to a statement that reads ‘Copyright (c) 2005, 2006, 2007, 2008,
 2009, 2010, 2011, 2012’."
+```
+
+<!-- links -->
+[bwa-mem.pl]: https://github.com/cancerit/PCAP-core/blob/master/bin/bwa_mem.pl
+[cgpmap-expected]: ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/expected
+[pcap-core]: https://github.com/cancerit/PCAP-core
+
+<!-- Travis -->
+[travis-base]: https://travis-ci.org/cancerit/dockstore-cgpmap
+[travis-master]: https://travis-ci.org/cancerit/dockstore-cgpmap.svg?branch=master
+[travis-develop]: https://travis-ci.org/cancerit/dockstore-cgpmap.svg?branch=develop
+
+<!-- Gitter -->
+[gitter-svg]: https://badges.gitter.im/dockstore-cgp/Lobby.svg
+[gitter-badge]: https://gitter.im/dockstore-cgp/Lobby?utm_source=badge&utm_medium=badge&utm_campaign=pr-badge&utm_content=badge
+
+<!-- Quay.io -->
+[quay-status]: https://quay.io/repository/wtsicgp/dockstore-cgpmap/status
+[quay-repo]: https://quay.io/repository/wtsicgp/dockstore-cgpmap
+[quay-builds]: https://quay.io/repository/wtsicgp/dockstore-cgpmap?tab=builds
+
+<!-- dockstore -->
+[dockstore-cgpmap]: https://dockstore.org/containers/quay.io/wtsicgp/dockstore-cgpmap
+[dockstore-get-started]: https://dockstore.org/docs/getting-started-with-dockstore
diff --git a/build/apt-build.sh b/build/apt-build.sh
deleted file mode 100755
index ba3b5fb..0000000
--- a/build/apt-build.sh
+++ /dev/null
@@ -1,18 +0,0 @@
-#!/bin/bash
-
-set -eux
-
-apt-get -yq update
-apt-get install -yq apt-transport-https
-apt-get install -yq --no-install-recommends curl
-apt-get install -yq --no-install-recommends build-essential
-apt-get install -yq --no-install-recommends libcurl4-openssl-dev
-apt-get install -yq --no-install-recommends nettle-dev
-apt-get install -yq --no-install-recommends libncurses5-dev
-apt-get install -yq autoconf
-apt-get install -yq --no-install-recommends libtool
-apt-get install -yq --no-install-recommends rsync
-apt-get install -yq --no-install-recommends libexpat1-dev
-apt-get install -yq --no-install-recommends time
-apt-get install -yq --no-install-recommends lsof
-apt-get install -yq --no-install-recommends libgoogle-perftools-dev
diff --git a/build/biobambam2-build.sh b/build/biobambam2-build.sh
deleted file mode 100644
index c625994..0000000
--- a/build/biobambam2-build.sh
+++ /dev/null
@@ -1,125 +0,0 @@
-#!/bin/bash
-LIBMAUSVERSION=2.0.312-release-20170208002133
-BIOBAMBAMVERSION=2.0.69-release-20170127133459
-SNAPPYVERSION=1.1.4
-IOLIBVERSION=1.14.8
-CHRPATHVERSION=0.16.1
-BUILDDIR=${PWD}
-INSTALLDIR=${BUILDDIR}/install-dir
-TOOLSDIR=${BUILDDIR}/tools-dir
-PAR=`cat /proc/cpuinfo | egrep "^processor" | wc -l`
-
-set -uxe
-
-# get chrpath
-if [ ! -f chrpath-${CHRPATHVERSION}.tar.gz ] ; then
-	curl -sSL "https://alioth.debian.org/frs/download.php/file/3979/chrpath-${CHRPATHVERSION}.tar.gz" > chrpath-${CHRPATHVERSION}.tar.gz
-fi
-
-# get iolib
-if [ ! -f io_lib-${IOLIBVERSION}.tar.gz ] ; then
-	curl -sSL "http://downloads.sourceforge.net/project/staden/io_lib/${IOLIBVERSION}/io_lib-${IOLIBVERSION}.tar.gz?&use_mirror=kent" \
-		> io_lib-${IOLIBVERSION}.tar.gz
-fi
-
-# get snappy
-if [ ! -f snappy-${SNAPPYVERSION}.tar.gz ] ; then
-  curl -sSL https://github.com/google/snappy/archive/${SNAPPYVERSION}.tar.gz > snappy-${SNAPPYVERSION}.tar.gz
-fi
-
-# get libmaus
-if [ ! -f libmaus-${LIBMAUSVERSION}.tar.gz ] ; then
-	curl -sSL https://github.com/gt1/libmaus2/archive/${LIBMAUSVERSION}.tar.gz > libmaus-${LIBMAUSVERSION}.tar.gz
-fi
-
-# get biobambam
-if [ ! -f biobambam-${BIOBAMBAMVERSION}.tar.gz ] ; then
-	curl -sSL https://github.com/gt1/biobambam2/archive/${BIOBAMBAMVERSION}.tar.gz > biobambam-${BIOBAMBAMVERSION}.tar.gz
-fi
-
-mkdir -p chrpath-${CHRPATHVERSION}-src
-tar -C chrpath-${CHRPATHVERSION}-src --strip-components=1 -xzf chrpath-${CHRPATHVERSION}.tar.gz
-mkdir -p chrpath-${CHRPATHVERSION}-build
-cd chrpath-${CHRPATHVERSION}-build
-${BUILDDIR}/chrpath-${CHRPATHVERSION}-src/configure --prefix=${TOOLSDIR}
-make -j${PAR}
-make -j${PAR} install
-cd ..
-rm -fR chrpath-${CHRPATHVERSION}-src chrpath-${CHRPATHVERSION}-build
-
-rm -fR ${INSTALLDIR}
-mkdir -p ${INSTALLDIR}
-
-# build iolib
-mkdir -p io_lib-${IOLIBVERSION}-src
-tar -C io_lib-${IOLIBVERSION}-src --strip-components=1 -xzf io_lib-${IOLIBVERSION}.tar.gz
-mkdir -p io_lib-${IOLIBVERSION}-build
-cd io_lib-${IOLIBVERSION}-build
-LDFLAGS="-Wl,-rpath=XORIGIN/../lib" ${BUILDDIR}/io_lib-${IOLIBVERSION}-src/configure --prefix=${INSTALLDIR}
-make -j${PAR}
-make -j${PAR} install
-cd ..
-rm -fR io_lib-${IOLIBVERSION}-src io_lib-${IOLIBVERSION}-build
-
-# build snappy
-mkdir -p snappy-${SNAPPYVERSION}-src
-tar -C snappy-${SNAPPYVERSION}-src --strip-components=1 -xzf snappy-${SNAPPYVERSION}.tar.gz
-cd snappy-${SNAPPYVERSION}-src
-autoreconf -i -f
-cd ../
-mkdir -p snappy-${SNAPPYVERSION}-build
-cd snappy-${SNAPPYVERSION}-build
-LDFLAGS="-Wl,-rpath=XORIGIN/../lib" ${BUILDDIR}/snappy-${SNAPPYVERSION}-src/configure --prefix=${INSTALLDIR}
-make -j${PAR}
-make -j${PAR} install
-cd ..
-rm -fR snappy-${SNAPPYVERSION}-src snappy-${SNAPPYVERSION}-build
-
-# build libmaus
-mkdir -p libmaus-${LIBMAUSVERSION}-src
-tar -C libmaus-${LIBMAUSVERSION}-src --strip-components=1 -xzf libmaus-${LIBMAUSVERSION}.tar.gz
-mkdir -p libmaus-${LIBMAUSVERSION}-build
-cd libmaus-${LIBMAUSVERSION}-build
-LDFLAGS="-Wl,-rpath=XORIGIN/../lib" ${BUILDDIR}/libmaus-${LIBMAUSVERSION}-src/configure --prefix=${INSTALLDIR} \
-	--with-snappy=${INSTALLDIR} \
-	--with-io_lib=${INSTALLDIR}
-make -j${PAR}
-make -j${PAR} install
-cd ..
-rm -fR libmaus-${LIBMAUSVERSION}-src libmaus-${LIBMAUSVERSION}-build
-
-# build biobambam
-mkdir -p biobambam-${BIOBAMBAMVERSION}-src
-tar -C biobambam-${BIOBAMBAMVERSION}-src --strip-components=1 -xzf biobambam-${BIOBAMBAMVERSION}.tar.gz
-mkdir -p biobambam-${BIOBAMBAMVERSION}-build
-cd biobambam-${BIOBAMBAMVERSION}-build
-LDFLAGS="-Wl,-rpath=XORIGIN/../lib" ${BUILDDIR}/biobambam-${BIOBAMBAMVERSION}-src/configure --prefix=${INSTALLDIR} \
-	--with-libmaus2=${INSTALLDIR}
-make -j${PAR}
-make -j${PAR} install
-cd ..
-rm -fR biobambam-${BIOBAMBAMVERSION}-src biobambam-${BIOBAMBAMVERSION}-build
-
-for i in `find ${INSTALLDIR} -name \*.so\*` ; do
-	ORIG=`objdump -x ${i} | grep RPATH | awk '{print $2}'`
-	MOD=`echo "$ORIG" | sed "s/XORIGIN/\\$ORIGIN/"`
-	${TOOLSDIR}/bin/chrpath -r "${MOD}" ${i}
-done
-
-for i in ${INSTALLDIR}/bin/* ; do
-	if [ ! -z `LANG=C file ${i} | egrep "ELF.*executable" | awk '{print $1}' | perl -p -e "s/://"` ] ; then
-		ORIG=`objdump -x ${i} | grep RPATH | awk '{print $2}'`
-		MOD=`echo "$ORIG" | sed "s/XORIGIN/\\$ORIGIN/"`
-		${TOOLSDIR}/bin/chrpath -r "${MOD}" ${i}
-	fi
-done
-
-rm -fR ${TOOLSDIR}
-
-# my additions
-mv ${INSTALLDIR} biobambam
-rm -f chrpath-${CHRPATHVERSION}.tar.gz
-rm -f io_lib-${IOLIBVERSION}.tar.gz
-rm -f snappy-${SNAPPYVERSION}.tar.gz
-rm -f libmaus-${LIBMAUSVERSION}.tar.gz
-rm -f biobambam-${BIOBAMBAMVERSION}.tar.gz
diff --git a/build/opt-build.sh b/build/opt-build.sh
index 5775372..aab9976 100755
--- a/build/opt-build.sh
+++ b/build/opt-build.sh
@@ -8,16 +8,17 @@ fi
 
 set -u
 
+VER_BBB2="2.0.86-release-20180228171821"
+
 ## for cgpBigWig
-VER_BIODBHTS="2.7"
-VER_LIBBW="0.3.1"
-VER_CGPBIGWIG="0.4.1"
+VER_BIODBHTS="2.9"
+VER_LIBBW="0.4.2"
 
 # for PCAP
-VER_BWA="v0.7.15"
-VER_HTSLIB="1.3.2"
-VER_SAMTOOLS="1.3.1"
-VER_PCAP="4.0.2"
+VER_BWA="v0.7.17"
+VER_HTSLIB="1.7"
+VER_SAMTOOLS="1.7"
+VER_PCAP="4.1.3"
 
 if [ "$#" -lt "1" ] ; then
   echo "Please provide an installation path such as /opt/ICGC"
@@ -48,43 +49,69 @@ else
 fi
 echo "Max compilation CPUs set to $CPU"
 
-
 SETUP_DIR=$INIT_DIR/install_tmp
 mkdir -p $SETUP_DIR/distro # don't delete the actual distro directory until the very end
 mkdir -p $INST_PATH/bin
 cd $SETUP_DIR
 
+## biobambam2 first
+BB_INST=$INST_PATH/biobambam2
+if [ ! -e $SETUP_DIR/bbb2.sucess ]; then
+  curl -sSL --retry 10 https://github.com/gt1/biobambam2/releases/download/${VER_BBB2}/biobambam2-${VER_BBB2}-x86_64-etch-linux-gnu.tar.gz > distro.tar.gz
+  mkdir -p $BB_INST
+  tar --strip-components 3 -C $BB_INST -zxf distro.tar.gz
+  rm -f $BB_INST/bin/curl # don't let this file in SSL doesn't work
+  rm -rf distro.* distro/*
+  touch $SETUP_DIR/bbb2.success
+fi
+
 # make sure tools installed can see the install loc of libraries
 set +u
 export LD_LIBRARY_PATH=`echo $INST_PATH/lib:$LD_LIBRARY_PATH | perl -pe 's/:\$//;'`
-export PATH=`echo $INST_PATH/bin:$PATH | perl -pe 's/:\$//;'`
-export MANPATH=`echo $INST_PATH/man:$INST_PATH/share/man:$MANPATH | perl -pe 's/:\$//;'`
+export PATH=`echo $INST_PATH/bin:$BB_INST/bin:$PATH | perl -pe 's/:\$//;'`
+export MANPATH=`echo $INST_PATH/man:$BB_INST/man:$INST_PATH/share/man:$MANPATH | perl -pe 's/:\$//;'`
 export PERL5LIB=`echo $INST_PATH/lib/perl5:$PERL5LIB | perl -pe 's/:\$//;'`
 set -u
 
 ## INSTALL CPANMINUS
 set -eux
 curl -sSL https://cpanmin.us/ > $SETUP_DIR/cpanm
-perl $SETUP_DIR/cpanm --no-interactive --notest --mirror http://cpan.metacpan.org -l $INST_PATH App::cpanminus
+perl $SETUP_DIR/cpanm --no-wget --no-interactive --notest --mirror http://cpan.metacpan.org -l $INST_PATH App::cpanminus
 rm -f $SETUP_DIR/cpanm
 
 ##### DEPS for cgpBigWig #####
 
 ## HTSLIB (tar.bz2)
 if [ ! -e $SETUP_DIR/htslib.success ]; then
+  rm -rf htslib
+  mkdir -p htslib
   curl -sSL --retry 10 https://github.com/samtools/htslib/releases/download/${VER_HTSLIB}/htslib-${VER_HTSLIB}.tar.bz2 > distro.tar.bz2
-  rm -rf distro/*
-  tar --strip-components 1 -C distro -jxf distro.tar.bz2
-  cd distro
+  tar --strip-components 1 -C htslib -jxf distro.tar.bz2
+  cd htslib
   ./configure --enable-plugins --enable-libcurl --prefix=$INST_PATH
   make clean
   make -j$CPU
   make install
   cd $SETUP_DIR
-  rm -rf distro.* distro/*
+  rm -rf distro.*
   touch $SETUP_DIR/htslib.success
 fi
 
+## SAMTOOLS (tar.bz2)
+if [ ! -e $SETUP_DIR/samtools.success ]; then
+  curl -sSL --retry 10 https://github.com/samtools/samtools/releases/download/${VER_SAMTOOLS}/samtools-${VER_SAMTOOLS}.tar.bz2 > distro.tar.bz2
+  rm -rf distro/*
+  tar --strip-components 1 -C distro -xjf distro.tar.bz2
+  cd distro
+  ./configure --enable-plugins --enable-libcurl --with-htslib=$INST_PATH --prefix=$INST_PATH
+  make clean
+  make -j$CPU all
+  make install
+  cd $SETUP_DIR
+  rm -rf distro.* distro/*
+  touch $SETUP_DIR/samtools.success
+fi
+
 ## LIB-BW (tar.gz)
 if [ ! -e $SETUP_DIR/libBigWig.success ]; then
   curl -sSL --retry 10 https://github.com/dpryan79/libBigWig/archive/${VER_LIBBW}.tar.gz > distro.tar.gz
@@ -96,24 +123,6 @@ if [ ! -e $SETUP_DIR/libBigWig.success ]; then
   touch $SETUP_DIR/libBigWig.success
 fi
 
-##### cgpBigWig installation
-if [ ! -e $SETUP_DIR/cgpBigWig.success ]; then
-  curl -sSL --retry 10 https://github.com/cancerit/cgpBigWig/archive/${VER_CGPBIGWIG}.tar.gz > distro.tar.gz
-  rm -rf distro/*
-  tar --strip-components 1 -C distro -xzf distro.tar.gz
-  make -C distro/c clean
-  make -C distro/c -j$CPU prefix=$INST_PATH HTSLIB=$INST_PATH/lib
-  cp distro/bin/bam2bedgraph $INST_PATH/bin/.
-  cp distro/bin/bwjoin $INST_PATH/bin/.
-  cp distro/bin/bam2bw $INST_PATH/bin/.
-  cp distro/bin/bwcat $INST_PATH/bin/.
-  cp distro/bin/bam2bwbases $INST_PATH/bin/.
-  cp distro/bin/bg2bw $INST_PATH/bin/.
-  cp distro/bin/detectExtremeDepth $INST_PATH/bin/.
-  rm -rf distro.* distro/*
-  touch $SETUP_DIR/cgpBigWig.success
-fi
-
 ##### DEPS for PCAP - layered on top #####
 
 ## build BWA (tar.gz)
@@ -127,73 +136,48 @@ if [ ! -e $SETUP_DIR/bwa.success ]; then
   touch $SETUP_DIR/bwa.success
 fi
 
-## SAMTOOLS (tar.bz2)
-if [ ! -e $SETUP_DIR/samtools.success ]; then
-  curl -sSL --retry 10 https://github.com/samtools/samtools/releases/download/${VER_SAMTOOLS}/samtools-${VER_SAMTOOLS}.tar.bz2 > distro.tar.bz2
-  rm -rf distro/*
-  tar --strip-components 1 -C distro -xjf distro.tar.bz2
-  cd distro
-  ./configure --enable-plugins --enable-libcurl --with-htslib=$INST_PATH --prefix=$INST_PATH
-  make clean
-  make -j$CPU all
-  make install
-  cd $SETUP_DIR
-  rm -rf distro.* distro/*
-  touch $SETUP_DIR/samtools.success
-fi
-
 ## Bio::DB::HTS (tar.gz)
 if [ ! -e $SETUP_DIR/Bio-DB-HTS.success ]; then
   ## add perl deps
-  cpanm --no-interactive --notest --mirror http://cpan.metacpan.org -l $INST_PATH Module::Build
-  cpanm --no-interactive --notest --mirror http://cpan.metacpan.org -l $INST_PATH Bio::Root::Version
+  cpanm --no-wget --no-interactive --notest --mirror http://cpan.metacpan.org -l $INST_PATH Module::Build
+  cpanm --no-wget --no-interactive --notest --mirror http://cpan.metacpan.org -l $INST_PATH Bio::Root::Version
 
   curl -sSL --retry 10 https://github.com/Ensembl/Bio-DB-HTS/archive/${VER_BIODBHTS}.tar.gz > distro.tar.gz
   rm -rf distro/*
   tar --strip-components 1 -C distro -zxf distro.tar.gz
   cd distro
   perl Build.PL --install_base=$INST_PATH --htslib=$INST_PATH
+  ./Build
+  ./Build test
   ./Build install
   cd $SETUP_DIR
   rm -rf distro.* distro/*
   touch $SETUP_DIR/Bio-DB-HTS.success
 fi
 
-## biobambam2
-if [ ! -e $SETUP_DIR/biobambam2.success ]; then
-  # co-located external script as it's complex
-  cd distro
-  rm -rf *
-  bash $SCRIPT_PATH/biobambam2-build.sh
-  cp biobambam/bin/* $INST_PATH/bin/.
-  rsync -rl biobambam/bin $INST_PATH/.
-  rsync -rl biobambam/include $INST_PATH/,
-  rsync -rl biobambam/lib $INST_PATH/.
-  rsync -rl biobambam/share $INST_PATH/.
-  cd $SETUP_DIR
-  rm -rf distro.* distro/*
-  touch $SETUP_DIR/biobambam2.success
-fi
-
 ##### PCAP-core installation
 
 if [ ! -e $SETUP_DIR/PCAP.success ]; then
-  cpanm --no-interactive --notest --mirror http://cpan.metacpan.org -l $INST_PATH Const::Fast
-  cpanm --no-interactive --notest --mirror http://cpan.metacpan.org -l $INST_PATH File::Which
+  cpanm --no-wget --no-interactive --notest --mirror http://cpan.metacpan.org -l $INST_PATH Const::Fast
+  cpanm --no-wget --no-interactive --notest --mirror http://cpan.metacpan.org -l $INST_PATH File::Which
   curl -sSL --retry 10 https://github.com/cancerit/PCAP-core/archive/${VER_PCAP}.tar.gz > distro.tar.gz
   rm -rf distro/*
   tar --strip-components 1 -C distro -xzf distro.tar.gz
   cd distro
   if [ ! -e $SETUP_DIR/pcap_c.success ]; then
     make -C c clean
-    make -C c -j$CPU prefix=$INST_PATH HTSLIB=$INST_PATH/lib
+    export REF_PATH=/tmp/REF_CACHE/cache/%2s/%2s/%s:http://www.ebi.ac.uk/ena/cram/md5/%s
+    export REF_CACHE=/tmp/REF_CACHE/cache/%2s/%2s/%s
+    mkdir -p /tmp/REF_CACHE
+    env HTSLIB=$SETUP_DIR/htslib make -C c -j$CPU prefix=$INST_PATH
     cp bin/bam_stats $INST_PATH/bin/.
     cp bin/reheadSQ $INST_PATH/bin/.
     cp bin/diff_bams $INST_PATH/bin/.
+    cp bin/mismatchQc $INST_PATH/bin/.
     touch $SETUP_DIR/pcap_c.success
   fi
-  cpanm --no-interactive --notest --mirror http://cpan.metacpan.org --notest -l $INST_PATH --installdeps .
-  cpanm -v --no-interactive --mirror http://cpan.metacpan.org -l $INST_PATH .
+  cpanm --no-wget --no-interactive --notest --mirror http://cpan.metacpan.org --notest -l $INST_PATH --installdeps .
+  cpanm -v --no-wget --no-interactive --mirror http://cpan.metacpan.org -l $INST_PATH .
   cd $SETUP_DIR
   rm -rf distro.* distro/*
   touch $SETUP_DIR/PCAP.success
@@ -201,17 +185,3 @@ fi
 
 cd $HOME
 rm -rf $SETUP_DIR
-
-set +x
-
-echo "
-################################################################
-
-  To use the non-central tools you need to set the following
-    export LD_LIBRARY_PATH=$INST_PATH/lib:\$LD_LIBRARY_PATH
-    export PATH=$INST_PATH/bin:\$PATH
-    export MANPATH=$INST_PATH/man:$INST_PATH/share/man:\$MANPATH
-    export PERL5LIB=$INST_PATH/lib/perl5:\$PERL5LIB
-
-################################################################
-"
diff --git a/build/perllib-build.sh b/build/perllib-build.sh
deleted file mode 100755
index 782620a..0000000
--- a/build/perllib-build.sh
+++ /dev/null
@@ -1,5 +0,0 @@
-#!/bin/bash
-
-set -eux
-
-echo "Doing nothing, here for perl libs that won't install through cpanminus"
diff --git a/cwls/cgpmap-bamBaiOut.cwl b/cwls/cgpmap-bamBaiOut.cwl
new file mode 100644
index 0000000..16814fe
--- /dev/null
+++ b/cwls/cgpmap-bamBaiOut.cwl
@@ -0,0 +1,115 @@
+#!/usr/bin/env cwl-runner
+
+class: CommandLineTool
+
+id: "cgpmap"
+
+label: "CGP BWA-mem mapping flow"
+
+cwlVersion: v1.0
+
+doc:
+  $include: includes/doc.yml
+
+requirements:
+  - $mixin: mixins/requirements.yml
+
+hints:
+  - $mixin: mixins/hints.yml
+
+inputs:
+  reference:
+    type: File
+    doc: "The core reference (fa, fai, dict) as tar.gz"
+    inputBinding:
+      prefix: -reference
+      position: 1
+      separate: true
+
+  bwa_idx:
+    type: File
+    doc: "The BWA indexes in tar.gz"
+    inputBinding:
+      prefix: -bwa_idx
+      position: 2
+      separate: true
+
+  sample:
+    type: string
+    doc: "Sample name to be included in output BAM header, also used to name final file"
+    inputBinding:
+      prefix: -sample
+      position: 3
+      separate: true
+
+  bwa:
+    type: string?
+    default: ' -Y -K 100000000'
+    doc: "Mapping and output parameters to pass to BWA-mem, see BWA docs, default ' -Y -K 100000000'"
+    inputBinding:
+      prefix: -bwa
+      position: 4
+      separate: true
+      shellQuote: false
+
+  groupinfo:
+    type: File?
+    doc: "Readgroup metadata file for FASTQ inputs"
+    inputBinding:
+      prefix: -groupinfo
+      position: 5
+      separate: true
+
+  mmqc:
+    type: boolean
+    doc: "Apply mismatch QC to reads following duplicate marking."
+    inputBinding:
+      prefix: -qc
+      position: 6
+
+  mmqcfrac:
+    type: float?
+    default: 0.05
+    doc: "Mismatch fraction to set as max before failing a read [0.05]"
+    inputBinding:
+      prefix: -qcf
+      position: 7
+      separate: true
+
+  seq_in:
+    type:
+    - 'null'
+    - type: array
+      items: File
+    doc: "Can be BAM, CRAM, fastq (paired or interleaved), BAM/CRAM can be mixed together but not FASTQ."
+    inputBinding:
+      position: 8
+
+outputs:
+  out_bam:
+    type: File
+    outputBinding:
+      glob: $(inputs.sample).bam
+    secondaryFiles:
+      - .bai
+      - .bas
+      - .md5
+      - .met
+      - .maptime
+
+baseCommand: ["/opt/wtsi-cgp/bin/ds-cgpmap.pl"]
+
+$schemas:
+  - http://schema.org/docs/schema_org_rdfa.html
+
+$namespaces:
+  s: http://schema.org/
+
+s:codeRepository: https://github.com/cancerit/dockstore-cgpmap
+s:license: https://spdx.org/licenses/AGPL-3.0-only
+
+s:author:
+  - class: s:Person
+    s:identifier: https://orcid.org/0000-0002-5634-1539
+    s:email: mailto:cgphelp@sanger.ac.uk
+    s:name: Keiran Raine
diff --git a/cwls/cgpmap-bamCsiOut.cwl b/cwls/cgpmap-bamCsiOut.cwl
new file mode 100644
index 0000000..934e261
--- /dev/null
+++ b/cwls/cgpmap-bamCsiOut.cwl
@@ -0,0 +1,115 @@
+#!/usr/bin/env cwl-runner
+
+class: CommandLineTool
+
+id: "cgpmap"
+
+label: "CGP BWA-mem mapping flow"
+
+cwlVersion: v1.0
+
+doc:
+  $include: includes/doc.yml
+
+requirements:
+  - $mixin: mixins/requirements.yml
+
+hints:
+  - $mixin: mixins/hints.yml
+
+inputs:
+  reference:
+    type: File
+    doc: "The core reference (fa, fai, dict) as tar.gz"
+    inputBinding:
+      prefix: -reference
+      position: 1
+      separate: true
+
+  bwa_idx:
+    type: File
+    doc: "The BWA indexes in tar.gz"
+    inputBinding:
+      prefix: -bwa_idx
+      position: 2
+      separate: true
+
+  sample:
+    type: string
+    doc: "Sample name to be included in output BAM header, also used to name final file"
+    inputBinding:
+      prefix: -sample
+      position: 3
+      separate: true
+
+  bwa:
+    type: string?
+    default: ' -Y -K 100000000'
+    doc: "Mapping and output parameters to pass to BWA-mem, see BWA docs, default ' -Y -K 100000000'"
+    inputBinding:
+      prefix: -bwa
+      position: 4
+      separate: true
+      shellQuote: false
+
+  groupinfo:
+    type: File?
+    doc: "Readgroup metadata file for FASTQ inputs"
+    inputBinding:
+      prefix: -groupinfo
+      position: 5
+      separate: true
+
+  mmqc:
+    type: boolean
+    doc: "Apply mismatch QC to reads following duplicate marking."
+    inputBinding:
+      prefix: -qc
+      position: 6
+
+  mmqcfrac:
+    type: float?
+    default: 0.05
+    doc: "Mismatch fraction to set as max before failing a read [0.05]"
+    inputBinding:
+      prefix: -qcf
+      position: 7
+      separate: true
+
+  seq_in:
+    type:
+    - 'null'
+    - type: array
+      items: File
+    doc: "Can be BAM, CRAM, fastq (paired or interleaved), BAM/CRAM can be mixed together but not FASTQ."
+    inputBinding:
+      position: 8
+
+outputs:
+  out_bam:
+    type: File
+    outputBinding:
+      glob: $(inputs.sample).bam
+    secondaryFiles:
+      - .csi
+      - .bas
+      - .md5
+      - .met
+      - .maptime
+
+baseCommand: ["/opt/wtsi-cgp/bin/ds-cgpmap.pl", "-csi"]
+
+$schemas:
+  - http://schema.org/docs/schema_org_rdfa.html
+
+$namespaces:
+  s: http://schema.org/
+
+s:codeRepository: https://github.com/cancerit/dockstore-cgpmap
+s:license: https://spdx.org/licenses/AGPL-3.0-only
+
+s:author:
+  - class: s:Person
+    s:identifier: https://orcid.org/0000-0002-5634-1539
+    s:email: mailto:cgphelp@sanger.ac.uk
+    s:name: Keiran Raine
diff --git a/cwls/cgpmap-cramOut.cwl b/cwls/cgpmap-cramOut.cwl
new file mode 100644
index 0000000..a59f8e4
--- /dev/null
+++ b/cwls/cgpmap-cramOut.cwl
@@ -0,0 +1,125 @@
+#!/usr/bin/env cwl-runner
+
+class: CommandLineTool
+
+id: "cgpmap"
+
+label: "CGP BWA-mem mapping flow"
+
+cwlVersion: v1.0
+
+doc:
+  $include: includes/doc.yml
+
+requirements:
+  - $mixin: mixins/requirements.yml
+
+hints:
+  - $mixin: mixins/hints.yml
+
+inputs:
+  reference:
+    type: File
+    doc: "The core reference (fa, fai, dict) as tar.gz"
+    inputBinding:
+      prefix: -reference
+      position: 1
+      separate: true
+
+  bwa_idx:
+    type: File
+    doc: "The BWA indexes in tar.gz"
+    inputBinding:
+      prefix: -bwa_idx
+      position: 2
+      separate: true
+
+  sample:
+    type: string
+    doc: "Sample name to be included in output CRAM header, also used to name final file"
+    inputBinding:
+      prefix: -sample
+      position: 3
+      separate: true
+
+  scramble:
+    type: string?
+    doc: "Options to pass to scramble when generating CRAM output, see scramble docs"
+    default: ''
+    inputBinding:
+      prefix: -scramble
+      position: 4
+      separate: true
+      shellQuote: true
+
+  bwa:
+    type: string?
+    default: ' -Y -K 100000000'
+    doc: "Mapping and output parameters to pass to BWA-mem, see BWA docs, default ' -Y -K 100000000'"
+    inputBinding:
+      prefix: -bwa
+      position: 5
+      separate: true
+      shellQuote: true
+
+  groupinfo:
+    type: File?
+    doc: "Readgroup metadata file for FASTQ inputs"
+    inputBinding:
+      prefix: -groupinfo
+      position: 6
+      separate: true
+
+  mmqc:
+    type: boolean
+    doc: "Apply mismatch QC to reads following duplicate marking."
+    inputBinding:
+      prefix: -qc
+      position: 7
+
+  mmqcfrac:
+    type: float?
+    default: 0.05
+    doc: "Mismatch fraction to set as max before failing a read [0.05]"
+    inputBinding:
+      prefix: -qcf
+      position: 8
+      separate: true
+
+  seq_in:
+    type:
+    - 'null'
+    - type: array
+      items: File
+    doc: "Can be BAM, CRAM, fastq (paired or interleaved), BAM/CRAM can be mixed together but not FASTQ."
+    inputBinding:
+      position: 9
+
+outputs:
+  out_cram:
+    type: File
+    outputBinding:
+      glob: $(inputs.sample).cram
+    secondaryFiles:
+      - .crai
+      - .bas
+      - .md5
+      - .met
+      - .maptime
+
+baseCommand: ["/opt/wtsi-cgp/bin/ds-cgpmap.pl", "-cram"]
+
+$schemas:
+  - http://schema.org/docs/schema_org_rdfa.html
+
+$namespaces:
+  s: http://schema.org/
+
+s:codeRepository: https://github.com/cancerit/dockstore-cgpmap
+s:license: https://spdx.org/licenses/AGPL-3.0-only
+
+s:author:
+  - class: s:Person
+    s:identifier: https://orcid.org/0000-0002-5634-1539
+    s:email: mailto:cgphelp@sanger.ac.uk
+    s:name: Keiran Raine
diff --git a/cwls/includes/doc.yml b/cwls/includes/doc.yml
new file mode 100644
index 0000000..2d25e2e
--- /dev/null
+++ b/cwls/includes/doc.yml
@@ -0,0 +1,9 @@
+![build_status](https://quay.io/repository/wtsicgp/dockstore-cgpmap/status)
+A Docker container for PCAP-core. See the [dockstore-cgpmap](https://github.com/cancerit/dockstore-cgpmap) website for more information.
+
+Please read the relevant [changes](https://github.com/cancerit/dockstore-cgpwgs/blob/master/CHANGES.md)
+when upgrading.
+
+Parameters for a CWL definition are generally described in a json file, but parameters can be provided on the command line.
+
+To see the parameters descriptions please run: cwltool --tool-help path_to.cwl
diff --git a/cwls/mixins/hints.yml b/cwls/mixins/hints.yml
new file mode 100644
index 0000000..0a08955
--- /dev/null
+++ b/cwls/mixins/hints.yml
@@ -0,0 +1,4 @@
+class: ResourceRequirement
+coresMin: 1 # works but long, 6 recommended
+ramMin: 15000 # good for WGS human ~30-60x
+outdirMin: 5000000 # unlikely any BAM processing would be possible in less
diff --git a/cwls/mixins/requirements.yml b/cwls/mixins/requirements.yml
new file mode 100644
index 0000000..ca87167
--- /dev/null
+++ b/cwls/mixins/requirements.yml
@@ -0,0 +1,2 @@
+class: DockerRequirement
+dockerPull: "quay.io/wtsicgp/dockstore-cgpmap:3.0.0"
diff --git a/examples/cgpmap/bamOut/bam_input.json b/examples/cgpmap/bamOut/bam_input.json
new file mode 100644
index 0000000..55368f7
--- /dev/null
+++ b/examples/cgpmap/bamOut/bam_input.json
@@ -0,0 +1,21 @@
+{
+  "reference": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/core_ref_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "bwa_idx": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/bwa_idx_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "seq_in": [
+    {"class": "File",
+      "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21.bam"}
+  ],
+  "sample": "test",
+  "mmqc": false,
+  "mmqcfrag": 0.05,
+  "out_bam": {
+    "path": "/tmp/mapped.bam",
+    "class": "File"
+  }
+}
diff --git a/examples/cgpmap/bamOut/fastq_gz_input.json b/examples/cgpmap/bamOut/fastq_gz_input.json
new file mode 100644
index 0000000..c878a89
--- /dev/null
+++ b/examples/cgpmap/bamOut/fastq_gz_input.json
@@ -0,0 +1,39 @@
+{
+  "reference": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/core_ref_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "bwa_idx": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/bwa_idx_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "seq_in": [
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10658_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10659_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10660_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10661_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10662_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10663_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10664_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10665_i.fq.gz"},
+  ],
+  "groupinfo": {
+    "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_fq.yaml",
+    "class": "File"
+  },
+  "sample": "test",
+  "mmqc": false,
+  "mmqcfrag": 0.05,
+  "out_bam": {
+    "path": "/tmp/mapped.bam",
+    "class": "File"
+  }
+}
diff --git a/examples/cgpmap/bamOutCsi/bam_input.json b/examples/cgpmap/bamOutCsi/bam_input.json
new file mode 100644
index 0000000..55368f7
--- /dev/null
+++ b/examples/cgpmap/bamOutCsi/bam_input.json
@@ -0,0 +1,21 @@
+{
+  "reference": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/core_ref_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "bwa_idx": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/bwa_idx_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "seq_in": [
+    {"class": "File",
+      "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21.bam"}
+  ],
+  "sample": "test",
+  "mmqc": false,
+  "mmqcfrag": 0.05,
+  "out_bam": {
+    "path": "/tmp/mapped.bam",
+    "class": "File"
+  }
+}
diff --git a/examples/cgpmap/bamOutCsi/fastq_gz_input.json b/examples/cgpmap/bamOutCsi/fastq_gz_input.json
new file mode 100644
index 0000000..c878a89
--- /dev/null
+++ b/examples/cgpmap/bamOutCsi/fastq_gz_input.json
@@ -0,0 +1,39 @@
+{
+  "reference": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/core_ref_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "bwa_idx": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/bwa_idx_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "seq_in": [
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10658_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10659_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10660_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10661_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10662_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10663_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10664_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10665_i.fq.gz"},
+  ],
+  "groupinfo": {
+    "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_fq.yaml",
+    "class": "File"
+  },
+  "sample": "test",
+  "mmqc": false,
+  "mmqcfrag": 0.05,
+  "out_bam": {
+    "path": "/tmp/mapped.bam",
+    "class": "File"
+  }
+}
diff --git a/examples/cgpmap/cramOut/bam_input.json b/examples/cgpmap/cramOut/bam_input.json
new file mode 100644
index 0000000..b16b605
--- /dev/null
+++ b/examples/cgpmap/cramOut/bam_input.json
@@ -0,0 +1,21 @@
+{
+  "reference": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/core_ref_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "bwa_idx": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/bwa_idx_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "seq_in": [
+    {"class": "File",
+      "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21.bam"}
+  ],
+  "sample": "test",
+  "mmqc": false,
+  "mmqcfrag": 0.05,
+  "out_cram": {
+    "path": "/tmp/mapped.cram",
+    "class": "File"
+  }
+}
diff --git a/examples/cgpmap/cramOut/fastq_gz_input.json b/examples/cgpmap/cramOut/fastq_gz_input.json
new file mode 100644
index 0000000..9504d95
--- /dev/null
+++ b/examples/cgpmap/cramOut/fastq_gz_input.json
@@ -0,0 +1,39 @@
+{
+  "reference": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/core_ref_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "bwa_idx": {
+    "path": "ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/human/bwa_idx_GRCh37d5.tar.gz",
+    "class": "File"
+  },
+  "seq_in": [
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10658_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10659_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10660_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10661_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10662_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10663_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10664_i.fq.gz"},
+    {"class": "File",
+     "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_10665_i.fq.gz"},
+  ],
+  "groupinfo": {
+    "path": "ftp://ngs.sanger.ac.uk/production/cancer/dockstore/cgpmap/insilico_21_fq.yaml",
+    "class": "File"
+  },
+  "sample": "test",
+  "mmqc": false,
+  "mmqcfrag": 0.05,
+  "out_cram": {
+    "path": "/tmp/mapped.cram",
+    "class": "File"
+  }
+}
diff --git a/examples/sample_configs.local.json b/examples/cgpmap/pre_3.0.json
similarity index 100%
rename from examples/sample_configs.local.json
rename to examples/cgpmap/pre_3.0.json
diff --git a/scripts/ds-cgpmap.pl b/scripts/ds-cgpmap.pl
new file mode 100755
index 0000000..25148d6
--- /dev/null
+++ b/scripts/ds-cgpmap.pl
@@ -0,0 +1,261 @@
+#!/usr/bin/perl
+
+use strict;
+use Getopt::Long;
+use File::Path qw(make_path);
+use Pod::Usage qw(pod2usage);
+use Data::Dumper;
+use autodie qw(:all);
+use warnings FATAL => 'all';
+
+pod2usage(-verbose => 1, -exitval => 1) if(@ARGV == 0);
+
+# set defaults
+my %opts = ('csi' => 0,
+            'c' => 0,
+            'q' => 0,
+            'sc' => q{},
+            'b' => q{},
+            'o' => $ENV{HOME},
+            't' => undef,
+            'g' => undef,
+            'f' => 0.05,
+            );
+
+GetOptions( 'h|help' => \$opts{'h'},
+            'm|man' => \$opts{'m'},
+            'r|reference=s' => \$opts{'r'},
+            'i|bwa_idx=s' => \$opts{'i'},
+            's|sample=s' => \$opts{'s'},
+            'c|cram' => \$opts{'c'},
+            'sc|scramble:s' => \$opts{'sc'},
+            'csi' => \$opts{'csi'},
+            'b|bwa:s' => \$opts{'b'},
+            'g|groupinfo:s' => \$opts{'g'},
+            't|threads:i' => \$opts{'t'},
+            'o|outdir:s' => \$opts{'o'},
+            'q|qc' => \$opts{'q'},
+            'f|qcf:f' => \$opts{'f'},
+) or pod2usage(2);
+
+pod2usage(-verbose => 1, -exitval => 0) if(defined $opts{'h'});
+pod2usage(-verbose => 2, -exitval => 0) if(defined $opts{'m'});
+
+delete $opts{'h'};
+delete $opts{'m'};
+
+printf "Options loaded: \n%s\n",Dumper(\%opts);
+
+# figure out if ref already unpacked:
+my $ref_area = $opts{'o'}.'/reference_files';
+my $ref_unpack = 1;
+if($opts{'r'} eq $opts{'i'} && -d $opts{'r'}) {
+  $ref_area = $opts{'r'};
+  $ref_unpack = 0;
+}
+
+# make the param file
+make_path($opts{'o'}) unless(-e $opts{'o'});
+my $run_file = $opts{'o'}.'/run.params';
+open my $FH,'>',$run_file or die "Failed to write to $run_file: $!";
+# Force explicit checking of file flush
+print $FH "export PCAP_THREADED_NO_SCRIPT=1\n";
+print $FH "export PCAP_THREADED_FORCE_SYNC=1\n";
+print $FH "export PCAP_THREADED_LOADBACKOFF=1\n";
+print $FH "export PCAP_THREADED_REM_LOGS=1\n";
+# General params
+printf $FH "REF_BASE='%s'\n", $ref_area;
+printf $FH "SAMPLE_NAME='%s'\n", $opts{'s'};
+printf $FH "OUTPUT_DIR='%s'\n", $opts{'o'};
+printf $FH "CRAM='%d'\n", $opts{'c'};
+printf $FH "SCRAMBLE='%s'\n", $opts{'sc'} if(length $opts{'sc'} > 0);
+printf $FH "CSI='%d'\n", $opts{'csi'};
+printf $FH "BWA_PARAM='%s'\n", $opts{'b'} if(length $opts{'b'} > 0);
+printf $FH "GROUPINFO='%s'\n", $opts{'g'} if(defined $opts{'g'});
+printf $FH "CPU=%d\n", $opts{'t'} if(defined $opts{'t'});
+printf $FH "CLEAN_REF=%d\n", $ref_unpack;
+printf $FH "INPUT='%s'\n", join ' ', @ARGV;
+printf $FH "MMQC=%d\n", $opts{'q'};
+printf $FH "MMQCFRAC=%s\n", $opts{'f'} if(defined $opts{'f'});
+close $FH;
+
+if($ref_unpack) {
+  ## unpack the reference area:
+  make_path($ref_area);
+  my $untar = sprintf 'tar --strip-components 1 -C %s -zxvf %s', $ref_area, $opts{'r'};
+  system($untar) && die $!;
+  $untar = sprintf 'tar --strip-components 1 -C %s -zxvf %s', $ref_area, $opts{'i'};
+  system($untar) && die $!;
+}
+
+exec('mapping.sh', $run_file); # I will never return to the perl code
+
+__END__
+
+
+=head1 NAME
+
+ds-cgpmap.pl - Generate the param file and execute mapping.sh (for dockstore)
+
+=head1 SYNOPSIS
+
+ds-cgpmap.pl [options] [file(s)...]
+
+  Required parameters:
+    -reference   -r   Path to core reference tar.gz
+                       - if already unpacked provide base directory
+                       - see `-m` for full details
+    -bwa_idx     -i   Path to bwa index tar.gz
+                       - if already unpacked provide base directory
+                       - see `-m` for full details
+    -sample      -s   Sample name to be applied to output file.
+
+  Optional parameters:
+    -cram        -c   Output cram, see '-sc'
+    -scramble    -sc  Single quoted string of parameters to pass to Scramble when '-c' used
+                      - '-I,-O' are used internally and should not be provided
+    -bwa         -b     Single quoted string of additional parameters to pass to BWA
+                         - '-t,-p,-R' are used internally and should not be provided
+    -groupinfo   -g   Readgroup metadata file for FASTQ inputs, values are not validated (yaml).
+    -threads     -t   Set the number of cpu/cores available [default all].
+    -outdir      -o   Set the output folder [$HOME]
+    -qc          -q   Apply mismatch QC to reads following duplicate marking
+    -qcf         -f   Mismatch fraction to set as max before failing a read [0.05]
+
+  Other:
+    -help        -h   Brief help message.
+    -man         -m   Full documentation.
+
+File list can be full file names or wildcard, e.g.
+
+=over 4
+
+=item mutiple BAM inputs
+
+ ds-cgpmap.pl [options] input/*.bam
+
+=item multiple paired fastq inputs
+
+ ds-cgpmap.pl [options] input/*_[12].fq[.gz]
+
+=item multiple interleaved paired fastq inputs
+
+ ds-cgpmap.pl [options] input/*.fq[.gz]
+
+=item mixture of BAM and CRAM
+
+ ds-cgpmap.pl [options] input/*.bam input/*.cram
+
+=back
+
+=head1 DESCRIPTION
+
+stuff
+
+=head1 OPTION DETAILS
+
+=over 4
+
+=item B<-reference> B<-bwa_idx>
+
+B<-reference> should point to a core_ref_XXXX.tar.gz
+
+B<-bwa_idx> should point to a bwa_idx_XXXX.tar.gz
+
+See ftp://ftp.sanger.ac.uk/pub/cancer/dockstore/
+
+If both are equal and directories will assume already unpacked as
+
+  mkdir ref_base
+  tar -C ref_base --strip-components 1 -zxvf core_ref_XXXX.tar.gz
+  tar -C ref_base --strip-components 1 -zxvf bwa_idx_XXXX.tar.gz
+
+B<-reference>
+
+=item B<-sample>
+
+Name to be applied to output files.  Special characters will not be magically fixed.
+
+
+=item B<-cram>
+
+Final output file will be a CRAM file instead of BAM.  To tune the the compression methods see then
+B<-scramble> option.
+
+=item B<-scramble>
+
+Single quoted string of parameters to pass to Scramble when '-c' used.  Please see the Scramble
+documentation for details.
+
+Please note: '-I,-O' are used internally and should not be provided.
+
+=item B<-csi>
+
+Generate csi index instead of bai when output is BAM.
+
+=item B<-bwa>
+
+Single quoted string of additional parameters to pass to BWA.  Please see the 'bwa mem'
+documentation for details.
+
+Please note: '-t,-p,-R' are used internally and should not be provided.
+
+=item B<-groupinfo>
+
+Readgroup information metadata file, please see the PCAP wiki for format:
+
+https://github.com/cancerit/PCAP-core/wiki/File-Formats-groupinfo.yaml
+
+=item B<-threads>
+
+Sets the number of cores to be used during processing.  Default to use all at appropriate
+points in analysis.
+
+Recommend increments of 6 once 6 is exceeded.
+
+=item B<-outdir>
+
+Set the output directory.  Defaults to $HOME.
+
+NOTE: Should B<NOT> be set when working with dockstore wrapper.
+
+=item B<-qc>
+
+Apply mismatch QC to read with a mismatch fraction higher than that specified in B<-qcf>.
+
+=item B<-qcf>
+
+When B<-qc> is set reads with a mismatch rate greater than this value to QC_FAIL (512/0x200).
+
+An auxilary tag is also set so these can be identified.
+
+=back
+
+=head2 INPUT FILE TYPES
+
+There are several types of file that the script is able to process.
+
+=over 4
+
+=item f[ast]q
+
+A standard uncompressed fastq file.  Requires a pair of inputs with standard suffix of '_1' and '_2'
+immediately prior to '.f[ast]q' or an interleaved f[ast]q file where read 1 and 2 are adjacent
+in the file.
+
+
+=item f[ast]q.gz
+
+As *.f[ast]q but compressed with gzip.
+
+=item bam
+
+Single lane BAM files, RG line is transfered to aligned files.  Also accepts multi lane BAM.
+
+=item cram
+
+Single lane BAM files, RG line is transfered to aligned files.  Also accepts multi lane CRAM.
+
+=back
+
+=cut
diff --git a/scripts/ds-wrapper.pl b/scripts/ds-wrapper.pl
deleted file mode 100755
index 1e4bebe..0000000
--- a/scripts/ds-wrapper.pl
+++ /dev/null
@@ -1,156 +0,0 @@
-#!/usr/bin/perl
-
-use strict;
-use Getopt::Long;
-use File::Path qw(make_path);
-use Pod::Usage qw(pod2usage);
-use Data::Dumper;
-use autodie qw(:all);
-use warnings FATAL => 'all';
-
-pod2usage(-verbose => 1, -exitval => 1) if(@ARGV == 0);
-
-# set defaults
-my %opts = ('c'=>0,
-            'sc' => q{},
-            'b' => q{}
-            ,);
-
-GetOptions( 'h|help' => \$opts{'h'},
-            'm|man' => \$opts{'m'},
-            'r|reference=s' => \$opts{'r'},
-            'i|bwa_idx=s' => \$opts{'i'},
-            's|sample=s' => \$opts{'s'},
-            'c|cram' => \$opts{'c'},
-            'sc|scramble:s' => \$opts{'sc'},
-            'b|bwa:s' => \$opts{'b'},
-) or pod2usage(2);
-
-pod2usage(-verbose => 1, -exitval => 0) if(defined $opts{'h'});
-pod2usage(-verbose => 2, -exitval => 0) if(defined $opts{'m'});
-
-delete $opts{'h'};
-delete $opts{'m'};
-
-printf "Options loaded: \n%s\n",Dumper(\%opts);
-
-## unpack the reference area:
-my $ref_area = $ENV{HOME}.'/reference_files';
-make_path($ref_area);
-my $untar = sprintf 'tar --strip-components 1 -C %s -zxvf %s', $ref_area, $opts{'r'};
-system($untar) && die $!;
-$untar = sprintf 'tar --strip-components 1 -C %s -zxvf %s', $ref_area, $opts{'i'};
-system($untar) && die $!;
-
-my $run_file = $ENV{HOME}.'/run.params';
-open my $FH,'>',$run_file or die "Failed to write to $run_file: $!";
-# Force explicit checking of file flush
-printf $FH "export PCAP_THREADED_NO_SCRIPT=1\n";
-printf $FH "export PCAP_THREADED_FORCE_SYNC=1\n";
-# General params
-printf $FH "REF_BASE='%s'\n", $ref_area;
-printf $FH "SAMPLE_NAME='%s'\n", $opts{'s'};
-printf $FH "OUTPUT_DIR='%s'\n", $ENV{HOME};
-printf $FH "CRAM='%s'\n", $opts{'c'};
-printf $FH "SCRAMBLE='%s'\n", $opts{'sc'} if(length $opts{'sc'} > 0);
-printf $FH "BWA_PARAM='%s'\n", $opts{'b'} if(length $opts{'b'} > 0);
-printf $FH "INPUT='%s'\n", join ' ', @ARGV;
-close $FH;
-
-exec('mapping.sh'); # I will never return to the perl code
-
-__END__
-
-
-=head1 NAME
-
-dh-wrapper.pl - Generate the param file and execute mapping.sh (for dockstore)
-
-=head1 SYNOPSIS
-
-dh-wrapper.pl [options] [file(s)...]
-
-  Required parameters:
-    -reference   -r   Path to core reference tar.gz
-    -bwa_idx     -i   Path to bwa index tar.gz
-    -sample      -s   Sample name to be applied to output file.
-
-  Optional parameters:
-    -cram        -c   Output cram, see '-sc'
-    -scramble    -sc  Single quoted string of parameters to pass to Scramble when '-c' used
-                      - '-I,-O' are used internally and should not be provided
-    -bwa         -b     Single quoted string of additional parameters to pass to BWA
-                         - '-t,-p,-R' are used internally and should not be provided
-
-  Other:
-    -help        -h   Brief help message.
-    -man         -m   Full documentation.
-
-File list can be full file names or wildcard, e.g.
-
-=over 4
-
-=item mutiple BAM inputs
-
- dh-wrapper.pl [options] input/*.bam
-
-=item multiple paired fastq inputs
-
- dh-wrapper.pl [options] input/*_[12].fq[.gz]
-
-=item multiple interleaved paired fastq inputs
-
- dh-wrapper.pl [options] input/*.fq[.gz]
-
-=item mixture of BAM and CRAM
-
- dh-wrapper.pl [options] input/*.bam input/*.cram
-
-=back
-
-=head1 DESCRIPTION
-
-stuff
-
-=head1 OPTION DETAILS
-
-=over 4
-
-=item B<-reference>
-
-Path to mapping tar.gz reference files
-
-=item B<-sample>
-
-Name to be applied to output files.  Special characters will not be magically fixed.
-
-=back
-
-=head2 INPUT FILE TYPES
-
-There are several types of file that the script is able to process.
-
-=over 4
-
-=item f[ast]q
-
-A standard uncompressed fastq file.  Requires a pair of inputs with standard suffix of '_1' and '_2'
-immediately prior to '.f[ast]q' or an interleaved f[ast]q file where read 1 and 2 are adjacent
-in the file.
-
-
-=item f[ast]q.gz
-
-As *.f[ast]q but compressed with gzip.
-
-=item bam
-
-Single lane BAM files, RG line is transfered to aligned files.  Also accepts multi lane BAM.
-
-=item cram
-
-Single lane BAM files, RG line is transfered to aligned files.  Also accepts multi lane CRAM.
-
-=back
-
-=cut
diff --git a/scripts/mapping.sh b/scripts/mapping.sh
index 9d1d45e..cf03f60 100755
--- a/scripts/mapping.sh
+++ b/scripts/mapping.sh
@@ -9,22 +9,24 @@ echo -e "\nStart workflow: `date`\n"
 declare -a PRE_EXEC
 declare -a POST_EXEC
 
-if [ -z ${PARAM_FILE+x} ] ; then
+if [[ $# -eq 1 ]] ; then
+  PARAM_FILE=$1
+elif [ -z ${PARAM_FILE+x} ] ; then
   PARAM_FILE=$HOME/run.params
 fi
+
 echo "Loading user options from: $PARAM_FILE"
 if [ ! -f $PARAM_FILE ]; then
   echo -e "\tERROR: file indicated by PARAM_FILE not found: $PARAM_FILE" 1>&2
   exit 1
 fi
 source $PARAM_FILE
-env
 
 if [ -z ${CPU+x} ]; then
   CPU=`grep -c ^processor /proc/cpuinfo`
 fi
 
-if [ -d $INPUT ] ; then
+if [ -d "$INPUT" ] ; then
   INPUT="$INPUT/*"
 fi
 
@@ -33,6 +35,8 @@ echo -e "\tSAMPLE_NAME : $SAMPLE_NAME"
 echo -e "\tINPUT : $INPUT"
 echo -e "\tREF_BASE : $REF_BASE"
 echo -e "\tCRAM : $CRAM"
+echo -e "\tCSI : $CSI"
+echo -e "\tMMQC : $MMQC"
 if [ -z ${SCRAMBLE+x} ]; then
   echo -e "\tSCRAMBLE : <NOTSET>"
 else
@@ -43,6 +47,11 @@ if [ -z ${BWA_PARAM+x} ]; then
 else
   echo -e "\tBWA_PARAM : $BWA_PARAM"
 fi
+if [ -z ${GROUPINFO+x} ]; then
+  echo -e "\tGROUPINFO : <NOTSET>"
+else
+  echo -e "\tGROUPINFO : $GROUPINFO"
+fi
 set +u
 
 if [ ${#PRE_EXEC[@]} -eq 0 ]; then
@@ -56,28 +65,17 @@ fi
 set -u
 mkdir -p $OUTPUT_DIR
 
-# run any pre-exec step before attempting to access BAMs
-# logically the pre-exec could be pulling them
-if [ ! -f $OUTPUT_DIR/pre-exec.done ]; then
-  echo -e "\nRun PRE_EXEC: `date`"
-
-  for i in "${PRE_EXEC[@]}"; do
-    set -x
-    $i
-    { set +x; } 2> /dev/null
-  done
-  touch $OUTPUT_DIR/pre-exec.done
-fi
+TIME_EXT="bam"
 
 ADD_ARGS=''
 if [ $CRAM -gt 0 ]; then
   ADD_ARGS="$ADD_ARGS -c"
+  TIME_EXT="cram"
   if [ ! -z ${SCRAMBLE+x} ]; then
     ADD_ARGS="$ADD_ARGS -sc ' $SCRAMBLE'";
   fi
 fi
 
-
 # use a different malloc library when cores for mapping are over 8
 if [ $CPU -gt 7 ]; then
   ADD_ARGS="$ADD_ARGS -l /usr/lib/libtcmalloc_minimal.so"
@@ -85,12 +83,30 @@ fi
 
 # if BWA_PARAM set
 if [ ! -z ${BWA_PARAM+x} ]; then
-  ADD_ARGS="$ADD_ARGS -b ' $BWA_PARAM'"
+  ADD_ARGS="$ADD_ARGS -b '$BWA_PARAM'"
+fi
+
+# if GROUPINFO set
+if [ ! -z ${GROUPINFO+x} ]; then
+  ADD_ARGS="$ADD_ARGS -g $GROUPINFO"
+fi
+
+# if CSI set
+if [ $CSI -gt 0 ]; then
+  ADD_ARGS="$ADD_ARGS --csi"
+fi
+
+# if GROUPINFO set
+if [ $MMQC -gt 0 ]; then
+  ADD_ARGS="$ADD_ARGS --mmqc"
+  if [ ! -z ${MMQCFRAC+x} ]; then
+    ADD_ARGS="$ADD_ARGS --mmqcfrac $MMQCFRAC"
+  fi
 fi
 
 # -f set to be unfeasibly large to prevent splitting of lane data.
 set -x
-bash -c "/usr/bin/time -f $TIME_FORMAT -o $OUTPUT_DIR/$SAMPLE_NAME.bam.maptime \
+bash -c "/usr/bin/time -f $TIME_FORMAT -o $OUTPUT_DIR/$SAMPLE_NAME.$TIME_EXT.maptime \
  bwa_mem.pl -o $OUTPUT_DIR \
  -r $REF_BASE/genome.fa \
  -s $SAMPLE_NAME \
@@ -101,12 +117,9 @@ bash -c "/usr/bin/time -f $TIME_FORMAT -o $OUTPUT_DIR/$SAMPLE_NAME.bam.maptime \
  $INPUT"
 { set +x; } 2> /dev/null
 
-# run any post-exec step
-echo -e "\nRun POST_EXEC: `date`"
-for i in "${POST_EXEC[@]}"; do
-  set -x
-  $i
-  { set +x; } 2> /dev/null
-done
+# cleanup reference area, see ds-cgpmap.pl
+if [ $CLEAN_REF -gt 0 ]; then
+  rm -rf $REF_BASE
+fi
 
 echo -e "\nWorkflow end: `date`"