Skip to main content

python module to reduce MUSE Raw data and combine them

Project description


WARNING: Doc need to be updated since many major changes. Coming asap.

This is a Python wrapper for the MUSE/VLT pipeline recipes. This runs using the esorex command lines (in parallel, using by default likwid, although other options are available). This wrapper is meant to reduce MUSE OBs automatically, after dumping the fits files in a given folder.

This package includes an alignment module which may be useful, independently given a set of images to align and assuming a reference image is provided. It can spit out an "OFFSET_LIST" MUSE compatible fits table including the flux scaling factors.

pymusepipe is also made for multi-pointing mosaics and multi-targets surveys as it will process targets automatically when provided a specific dictionary of which target and which pointings to consider.

1 - Getting Started

Here are some basic instructions to get you going with pymusepipe.


pymusepipe uses Python 3. It is not compatible with python 2.7. If you are still using python 2.7 I would recommend you to switch asap as it won't be supported much longer anyway.

pymusepipe needs a number of standard modules/packages including:

  • numpy
  • scipy for the alignment module
  • matplotlib if you wish to make some plots (e.g. via the check_pipe module)
  • astropy as a general provider of many useful functions and structures, most importantly the astropy Tables.

The convolution package of pymusepipe (allowing full convolution of a datacube) uses pypher which is thus a requirement for that module to work.

This also includes a heavy-based usage of mpdaf, which is a super useful package developed by the MUSE GTO-CRAL Team, to process and analyse datacubes, and more specifically MUSE cubes, images and spectra.


You can install this package via Pypi via a simple:

pip install pymusepipe

You can obviously also install it by cloning it from github, or downloading the source (from github) and do something like:

python develop

I advise to use the "develop" option as it actually does not copy the files in your system but just create a link. In that way you can easily update the source software without reinstalling it. The link will directly use the source which has been udpated.

The other option is to use the standard "install" option:

python install

2 - Basic Usage - Dealing with OBs, individually

The pymusepipe wrapper is meant to provide the user with a simple way to run the MUSE pipeline.

Only 3 steps are needed: preparing the data (download), preparing the configuration files (templates are provided) which should take a couple of minutes, and using the code (a few lines at most). The package has many other hidden functionalities which I'll document as soon as I can.

I recommend to use Ipython as an environment, possibly via a screen which would allow you to disconnect from the server that actually runs the commands. Jupyter notebooks may also be very handy although I would not necessarily recommend them for this specific purpose as the running of the pipeline may take hours.

Step 1 = Preparing your data

The first thing to do is to prepare the folder structure to welcome your MUSE datasets. This is actually rather simple.

Imagine you have

  • a target or field named "NGC1000" (just for the sake of using an example).
  • several pointings (each pointing can represent a set of OBs which have roughly the same coordinates, best is that they have a very significant overlap): 1, 2, and 3. Note that this 'pointing' structure may not reflect your own observational strategy. There are cases where one OB includes different pointings. What is important here is that each 'pointing' is either a full single OB, or several OBs with the same pointing.

Then under your data root folder <my_data_folder> create the following folder structure:


Each pointing/OB has a "Raw" folder.

The next step is to download your MUSE data from the ESO web site, and put all the raw files (fits/fits.gz) into each individual "Raw" folder, associated with the right pointing.

Step 2 = Preparing your configuration files

pymusepipe only needs 2 configurations ASCII files: 1 for the calibration (calib_tables.dic) and one for the root folders (rc.dic). Examples of such files are provided in the "config_templates" folder of the pymusepipe package.

Step 3 = Running the pipeline

The pipeline is meant to be run automatically from a given python structure. This should thus take the user only a few lines of codes, including one to import pymusepipe, a couple to define the names of the configuration files, one to initialise the musepipe python structure, and one to launch the recipes. Here is an example of how this should look:

# Import the modules
import pymusepipe as pmp
from pymusepipe import musepipe

# define the names of the configuration files
rcfile = "/my_data/MUSE/Config/rc.dic"
calfile = "my_data_MUSE/Config/calib_tables.dic"

# Initialisation of the python - MusePipe Class - structure
mypipe = musepipe.MusePipe(targetname="NGCXXXX", pointing=1, rc_filename=rcfile,
                          cal_filename=calfile, log_filename="NGCXXXX_version01.log",
                          fakemode=False, overwrite_astropy_table=True, 

# Launching the pipeline

Some explanation may be needed to understand what is happening:

  • "targetname": is just the name of the target, used to decided where the data will be
  • "pointing": will be used as "P01" for pointing=1, etc.
  • "logfile": name of the logging file. Note that this logfile is actually a shell-like file which can be used to re-run the pipeline one step at a time. Note that there will be also 2 more files created using that name <logfile_name>, namely: a file named <logfile_name>.out and <logfile_name>.err which will contain the full output of the commands, and the error messages (stdout, stderr, resp.).
  • "fakemode": you can set this to True if you just wish to initialise things without actually running any recipes. The pipeline will only set things up but if you run any recipes will only "fake" them (not launch any esorex command, only spitting the code out)
  • "filter_list": list of filter names to use to reconstruct images when building up cubes. This should be part of the filter_list fits table provided (see calib_tables config file).
  • "filter_for_alignment": specific filter name used for alignment between exposures.

Other options can be useful:

  • "musemode": this is by default "WFM_NOAO_N" which is the most often used MUSE mode. This will filter out exposures not compatible with the given mode. So please beware.
  • "reset_log": will reset the log file. By default it is False, hence new runs will be appended.
  • "overwrite_astropy_table": by default this is False. If True, new runs will rewrite the Astropy output tables.
  • "time_astrometry": by default it is False, meaning the pipeline will try to detect a GEOMETRY and ASTROMETRY Files delivered with the Raw files by ESO. If set to True, it will use the time dependent astro/geo files provided by the GTO Team but you would need to make these available on your system. Hence I would recommend to keep the default (False).

3- Advanced Usage - Targets and Mosaics

4- Alignment module

Details to come.

5- Further details

Here we provide a bit more details about a few issues regarding the configuration files or set up parameters.

Configuration files

rc configuration file

It contains 3 lines, with: musecalib, musecalib_time and root.

  • root provides the root folder for your data. For Target NGCXXXX, and Pointing 1, the Raw data will be looked for in root/TargetXXXX/P01/Raw (see Preparing your data above).
  • musecalib should contain the standard MUSE calibration files. These are usually distributed in the MUSE kit within a "muse-calib-x.x.x/cal" folder.
  • musecalib_time: time dependent geometry and astrometry files (the correspondence between observing run dates and specific files are given in the dic_geo_astrowcs_table in

calib_tables configuration file

It contains a series of given fits files which will be used by the pipeline. Most names are self-explantory. That includes:

  • geo_table and astro_table: only used if you don't rely on the default time dependent geometry files (see rc file).
  • badpix_table, vignetting_mask, std_flux_table, extinct_table, line_catalog all usually provided with the MUSE pipeline.
  • filter_list : name of the fits filter list. This is used in case you wish to provide your own. Note that it needs to follow the MUSE standard for such a table.


Most MUSE pipeline recipes are run while run_all_recipes is launched. This can be changed in the or just scripted. Actually run_all_recipes() is just a function which launches (assuming 'mypipe' is your MusePipe structure, see above):

        mypipe.run_scipost(expotype="SKY", offset_list=False, skymethod='none')

where "illum" is a boolean (default is True), fraction is 0.8.

Feel free to launch these steps one by one (the order is important as in any data reduction process).

Structure of the output


The structure of the output is driven by a set of folder names described in in a few dictionaries (dic_input_folders, dic_folders, dic_folders_target). You can in principle change the names of the folders themselves, but I would advise against that.

The pipeline will create the folder structure automatically, checking whether the folders exist or not.

Log files

2 basic log files are spitted out: one is the Esorex output which will be stored in the "Esorex_log" folder. The other one will be in the "Log" folder with the name provided at start: that one is like a shell script which can be used to rerun things directly via the command line. In the "Log" folder, there will also be, for each log file, a file ".out" and one with ".err" extensions, respectively including all the stdout and stderr messages. This may be useful to trace details in the data reduction and problems.

Astropy Tables

Each recipe will trigger the creation of a astropy Table. These are stored under "Astro_Tables". You can use these to monitor which files have been processed or used.

Python structure

Most of the information you may need is actually stored in the python "MusePipe" class structure. More details to come.


  • Eric Emsellem [2017-2020], at ESO and CRAL


This project is licensed under the MIT License - see the LICENSE file for details


I would like to thank people who have initially sent me their code-samples, including Bernd Husemann, Dimitri Gadotti, Lodovico Coccato, Mark den Brok. I would also like to specifically and warmly thank Rebecca McElroy who supported me with the MUSE data reduction at the early stages of the development of this package, and Francesco Santoro who has significantly contributed in the testing, debugging of the code (version 1), and proposed dedicated python lines to be integrated in pymusepipe (e.g., alignment module).

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pymusepipe-2.14.2.tar.gz (132.9 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page