Note that this pipeline requires Python 3.9 or above

PJPipe (the PHANGS-JWST-Pipeline) is a wrapper around the official STScI JWST Pipeline for imaging data (not spectroscopy), with edits specific to the reduction of large mosaics and nearby galaxies with extended, diffuse emission.

Beyond the standard pipeline, PJPipe offers options for

• NIRCam destriping
• Dealing with the MIRI coronagraph
• Background matching
• Absolute astrometric correction

Alongside this, PJPipe is also highly parallelised for speed, and provides a simple, high-level interface via configuration files.

If you make use of PJPipe in your work, please cite the PHANGS-JWST survey papers (Lee et al., 2022; Williams et al., in prep.), and do not hesitate to get in touch for help! The /config directory on the GitHub repository has examples, but different datasets may need more specific tailoring. You can open an issue if you run into problems.

The easiest way to install PJPipe is via pip:

pip install pjpipe

You must also install spacepylot for PJPipe to work. PyPI does not work with direct dependencies, so install this afterwards like:

pip install git+https://github.com/ejwatkins-astro/spacepylot.git

The pipeline is primarily interfaced with using config files. These are .toml, and as such are relatively human-readable. We separate out parameters that control the overall pipeline processing (config files) and ones that are specific to the system directory layout (local files).

At the very least, you should determine a list of targets:

targets = [
    'ic5332',
    'ngc0628',
    'ngc1087',
    'etc',
]

a version:

version = 'v0p8p2'

a list of bands

bands = ['F300M']

and some steps

steps = [
    'download',
    'lv1',
    'lv2',
    'single_tile_destripe.nircam',
    'get_wcs_adjust',
    'apply_wcs_adjust',
    'lyot_separate.miri',
    'multi_tile_destripe.nircam',
    'level_match',
    'lv3',
    'astrometric_catalog.miri',
    'astrometric_align',
    'release',
    'regress_against_previous',
]

Note that in the steps here, you can separate things out per-instrument. For instance, destriping only runs on NIRCam images, whilst anything to do with the lyot coronagraph only applies to MIRI.

There is also the option to separately image background observations. For this, append _bgr to a band:

bands = [
    'F2100W',
    'F2100W_bgr',
]

and you can also run steps separately depending on whether the observations are background or not:

steps = [
    'lyot_separate.miri.sci',
    'lyot_mask.miri.bgr',
]

This config file controls the parameters for each step. You can edit things like so:

[parameters.download]

prop_id = '2107'
product_type = [
    'SCIENCE',
]
calib_level = [
    1,
]

which will download data from Program ID 2107 (the PHANGS Cycle 1 Treasury), and only level 1 (uncal) files. For more examples, we suggest looking in the config/ directory. For any parameters passed to the JWST pipeline itself, these should be nested as jwst_parameters, e.g.:

[parameters.lv1]

jwst_parameters.save_results = true
jwst_parameters.ramp_fit.suppress_one_group = false
jwst_parameters.refpix.use_side_ref_pixels = true

The local.toml file simply defines where things will be saved. For example,

crds_path = '/data/beegfs/astro-storage/groups/schinnerer/williams/crds/'
raw_dir = '/data/beegfs/astro-storage/groups/schinnerer/williams/jwst_raw/archive_20230711/'
reprocess_dir = '/data/beegfs/astro-storage/groups/schinnerer/williams/jwst_phangs_reprocessed/'
alignment_dir = '/data/beegfs/astro-storage/groups/schinnerer/williams/jwst_scripts/examples/2107/alignment/'
processors = 20

This should be edited to match your system layout.

After you have your config files set up (see the /config directory for some examples), you can run the pipeline end-to-end with just a few lines:

from pjpipe import PJPipeline

config_file = '/path/to/config.toml'
local_file = '/path/to/local.toml'

pjp = PJPipeline(config_file=config_file,
                 local_file=local_file,
                 )
pjp.do_pipeline()

Then just sit back and enjoy the heavy lifting being done.

If this is your first time running anything JWST related, errors can occur because the pipeline expects some reference files. To fix this, after setting CRDS parameters run

import os

os.system('crds sync --jwst')

and this will pull the minimum relevant files for you.

Each step is highly configurable to the end user, although the defaults should be sensible in many use cases. To see all steps, you can simply do:

import pjpipe

pjpipe.list_steps()

To see the possible arguments for each step, you can do:

from pjpipe import DownloadStep

help(DownloadStep)

This will list the optional arguments you can put into the config file, and will be passed to the step. This is not necessarily true for each JWST pipeline step, for those we recommend looking at the online docs

• If you'd like to generate convolution kernels easily, this repository will do that for you

PJPipe has been developed by the PHANGS Team, with major contributions from:

• Thomas Williams (University of Oxford)
• Oleg Egorov (Universität Heidelberg)
• Erik Rosolowsky (University of Alberta)
• Francesco Belfiore (INAF)
• Jessica Sutter (UCSD)
• David Thilker (JHU)
• Adam Leroy (OSU)

The PCA implementation for NIRCam destriping was adapted from work by Elizabeth Watkins (University of Manchester), and up-to-date code is available at this Github repository.