parch

Command-line tools for the PARCH water-shell workflow: extract a solute from an equilibrated system, build a hydration shell for annealing, submit the annealing runs, analyse the per-residue dehydration, and compute averaged PARCH values.

The full pipeline is five commands:

parch prep  ->  parch shellsetup  ->  parch submit  ->  parch analysis  ->  parch calpv

For a whole lipid membrane, swap parch shellsetup for parch membshell (section 2b): same shell-carving idea, but it builds a non-cubic box that keeps the membrane's periodic XY size, lays counterions in planes above and below the slab rather than on a sphere, and can delete the waters that solvation buries inside the membrane core. The rest of the pipeline is unchanged.

Two auxiliary commands share the same inputs/outputs as their counterparts and are documented at the end: parch analysis_na (the sort-then-sample hydration trend, kept only to reproduce the previous DNA/RNA work) and parch cval (raw correlation integrals, before the hard_ref normalisation that turns them into PARCH values).

Installation

Prerequisites

Requirement	Needed for	Notes
Python ≥ 3.8 with pip	everything	a conda env is recommended
numpy, MDAnalysis ≥ 2.0	everything	installed automatically by pip
GROMACS (gmx)	shellsetup, membshell, and running the annealing	external; not installable via pip. Put it on PATH (e.g. source /path/to/gromacs/bin/GMXRC)
SLURM (sbatch)	submit only	external; only needed to launch annealing jobs on a cluster

prep, analysis and calpv are pure Python/MDAnalysis and need neither GROMACS nor SLURM.

Install from GitHub

Install straight from the repository into the Python environment where you want the parch commands to live:

python -m pip install git+https://github.com/XuyangTsin/parchpackage.git

or from a local clone:

git clone https://github.com/XuyangTsin/parchpackage.git
cd parch
python -m pip install .          # add -e for an editable/dev install

The bundled data (parch/backup_annealing/ — the mdp/itp/INP/evp templates — and parch/resblocks/ — the -blocks molecule-type templates) is shipped as package data, so a normal install includes everything the tools need at runtime — no editable install required. (Force fields are not bundled; use your own for the upstream equilibration / topology.) This puts the single parch umbrella command on the environment's PATH:

Command	Purpose
parch prep	extract selected molecules from an equilibrated system, and prepare files
parch shellsetup	build a hydration shell (+ counterions) and set up the box
parch membshell	shellsetup for a whole lipid membrane: planar ion layers, non-cubic box, intra-core water removal
parch submit	stage simulation files into mid_* and submit jobs
parch analysis	analyse the per-residue water rentation over the annealing (sample-then-sort trend; default)
parch calpv	calculate PARCH values per-residue or per-block, averaged over mid_* runs
parch analysis_na	same as analysis but the sort-then-sample dehydration trend, for DNA/RNA reproduction only (auxiliary)
parch cval	raw integrated correlation values (pre-hard_ref), averaged over mid_* runs (auxiliary)

Verify the install:

parch -h
parch shellsetup -h

All commands can be run from any directory; paths are resolved against where you launch them.

Cluster configuration (important for parch submit)

parch submit copies SLURM job scripts (evp_<i>_zest3.sh for cpu, evp_<i>_gpu.sh for gpu) into each run directory and submits them. These scripts are written for the original author's cluster — partitions (--partition=normal,longjobs), node excludes, and module load gromacs / mpirun gmx_mpi.

Before using parch submit on your own cluster,

edit the#SBATCH headers and the module/mpirun lines in the installed parch/backup_annealing/evp_*.sh templates

(runpython -c "import parch, os; print(os.path.join(os.path.dirname(parch.__file__), 'backup_annealing'))" to find them) to match your scheduler, partitions, and GROMACS build.

(You can also skip submit entirely and launch the shell/mid_* runs with your own scripts; each just needs gmx grompp/mdrun on W_init.gro, W_topol.top, W_ind.ndx with em.mdp then heat_nvt_<i>.mdp.)

Layout

parch/ is a proper Python package; its modules (prep, shellsetup, submit, analysis, calpv, cli) and its backup_annealing/ and resblocks/ data directories are installed together, so the tools find their templates relative to the installed package on any machine.

---

1. parch prep — extract the solute

This tool is useful for preparing the structure (.gro) and topology (.top) files before setting up the shell. It is most useful in interactive mode, but a non-interactive mode has also been implemented to support automated workflows and scripting.

From an equilibrated system (md.gro + md.tpr + .top), list every molecule type / chain, choose which to keep, and write a trimmed structure + topology ready for shellsetup. The molecule→atom mapping comes from the .tpr; the coordinates come from the .gro.

In addition to trimming the structure/topology, prep now also:

• makes the kept molecules whole across the periodic boundary and centres them in the box (using bonds from the .tpr), so the solute is never split when shellsetup carves the shell — no manual gmx trjconv -pbc step needed;
• lets you pick the counter-ion force field (-ionff charmm|amber) so the correct PARCH ion .itp files are included;
• can interactively build the shellsetup -shelldef file (per-molecule shell thicknesses) and the analysis -blocks file (e.g. DNA/RNA sugar/base splits) for you, using the renumbered residue ranges it just produced.

IMPORTANT: Before running the setup command, ensure that all heavy atoms of the molecules you want to run with PARCH have position restraints defined in their corresponding .itp files.

The position restraints should be enclosed by the POSRES preprocessor directive, for example:

#ifdef POSRES
[ position_restraints ]
    1     1    POSRES_FC_BB    POSRES_FC_BB    POSRES_FC_BB   
    5     1    POSRES_FC_BB    POSRES_FC_BB    POSRES_FC_BB   
    7     1    POSRES_FC_SC    POSRES_FC_SC    POSRES_FC_SC   
   10     1    POSRES_FC_SC    POSRES_FC_SC    POSRES_FC_SC   
   11     1    POSRES_FC_SC    POSRES_FC_SC    POSRES_FC_SC   
   12     1    POSRES_FC_SC    POSRES_FC_SC    POSRES_FC_SC   
   13     1    POSRES_FC_BB    POSRES_FC_BB    POSRES_FC_BB   
   14     1    POSRES_FC_SC    POSRES_FC_SC    POSRES_FC_SC   
   15     1    POSRES_FC_SC    POSRES_FC_SC    POSRES_FC_SC
   .......
   .......  
#endif

In default, these position restraints are activated in the mdp options for later annealing process as:

define                  = -DPOSRES -DPOSRES_FC_BB=10000 -DPOSRES_FC_SC=10000

POSRES_FC_BB and POSRES_FC_SC correspond to the restraint force constants (in kJ mol⁻¹ nm⁻²) applied to backbone and side-chain heavy atoms, respectively.

parch prep -f md.gro -s md.tpr -pi topol.top -o solute.gro -po solute.top

!!Please Read the NOTICE printed when excuting the command for guidance!!

First, it prints a numbered table of molecules and asks which to KEEP (keyboard, e.g. 1 2 4 5-10),

#group  molecule        num_mol  atoms
1       PROA            1        1418
2       HEME            1        73
3       SOD             88       88
4       CLA             77       77
5       TIP3            27337    82011

(It first prints a reminder not to keep the waters or equilibration ions here — e.g. SOL, TP3, OPC, NA, CL, K, CA — since shellsetup rebuilds the solvent shell and PARCH ships its own ion topologies.)

IMPORTANT: After selecting, it modifies the .top file, and reports the renumbered residue ranges (Consecutive, and starting from 1) in the output (can be used for -shelldef later):

NOTICE: Output residue ranges (renumbered) -- useful for -separateshell or -shelldef:
    PROA           1:90
    HEME           91:91

Then, it prints #include (force-field) lines, and asks which to keep (keyboard).

DO NOT KEEP the .itp files for IONS used during equilibration, as PARCH uses its own tailored ones. Keeping the original ion .itp files may result in conflicting topology entries.

If you plan to use a DIFFERENT water model than the one used during equilibration, DO NOT keep the water's .itp file.

The following itp files are in your .top:
1     toppar/forcefield.itp
2     toppar/PROA.itp
3     toppar/HEME.itp
4     toppar/SOD.itp
5     toppar/CLA.itp
6     /home/xqin10/parch_platform/backup_eqb/charmm36-jul2021.ff/tip3p.itp

After selection, the kept .itp files will be overwritten with their full path in the .top, to avoid additional copying and moving the files:

; Include forcefield parameters
#include "/home/xqin10/parch_platform/test_shellsetup/uniform/toppar/forcefield.itp"
#include "/home/xqin10/parch_platform/test_shellsetup/uniform/toppar/PROA.itp"
#include "/home/xqin10/parch_platform/backup_eqb/charmm36-jul2021.ff/tip3p.itp"

Additionally, PARCH's own tailored counter-ion .itp files are appended automatically. Which ones depends on the counter-ion force field: prep asks Which forcefield are you using? (charmm/amber) (or take it from -ionff). For CHARMM it adds CLA.itp + SOD.itp; for AMBER it adds Cl-_AMBER.itp + Na+_AMBER.itp:

; PARCH counter-ion topologies
#include "/home/xqin10/parch_platform/parch_package/backup_annealing/CLA.itp"
#include "/home/xqin10/parch_platform/parch_package/backup_annealing/SOD.itp"

IMPORTANT: The net charge of the retained molecules will be reported at the end. This value should be provided to the shellsetup to properly place counterions.

Building -shelldef / -blocks interactively

After writing the topology, prep offers to generate the two helper files that shellsetup and analysis consume, using the renumbered residue ranges it just printed (so the ranges are guaranteed to line up):

• -shelldef file (prompted, or forced with -separateshell yes|no): it lists the kept molecule groups, asks which are protein (auto 4.15 Å) and which are DNA/RNA (auto 4.50 Å), then lets you add any number of extra groups with a custom thickness. The result is written to -shelldefout (default shelldef.txt) for use with parch shellsetup -separateshell yes -shelldef shelldef.txt.
• -blocks file (prompted, or forced with -makeblocks yes|no): for each molecule-type template found in -resblocksdir (bundled: dna.txt, rna.txt, meth_dna.txt — add more by dropping in another <moltype>.txt), it asks which kept groups it applies to and expands the template's block/atom definitions onto those residue ranges. The result is written to -blocksout (default blocks.txt) for use with parch analysis -blocks blocks.txt.

Both steps are optional — press Enter to skip a selection — and are skipped entirely if you answer no.

When run non-interactively, pass -separateshell / -makeblocks (and, if needed, -ionff) so prep never blocks on a prompt.

Finally, prep prints a short checklist to verify before shellsetup: kept residues are sequentially numbered, molecules are centred and unbroken in the box, and any shelldef.txt/blocks.txt are correct.

Option	Meaning
-f	input equilibrated structure (md.gro)
-s	run input file (md.tpr) — supplies the molecule typing
-pi	input topology (.top)
-o	output structure (.gro) with kept molecules
-po	output topology (.top) with kept molecules / includes
-renumber	yes (default) renumbers kept residues from 1 (shellsetup-ready);
no keeps original ids, the structure will not work well for shell setup
-keep	non-interactive selection, e.g. "1 2 4" or "1-3,5"
-keepff	non-interactive #include selection (default with -keep: keep all)
-ionff	counter-ion force field for the added PARCH ion .itps: charmm (CLA/SOD) or amber (Cl-/Na+); prompted if omitted
-separateshell	yes/no — whether to interactively build a -shelldef file; prompted if omitted
-shelldefout	output shell-definition file written when -separateshell yes (default shelldef.txt)
-makeblocks	yes/no — whether to interactively build an analysis -blocks file; prompted if omitted
-blocksout	output block-mapping file written when -makeblocks yes (default blocks.txt)
-resblocksdir	directory of block templates, one <moltype>.txt per type (default: bundled resblocks/)

---

2. parch shellsetup — build the hydration shell and place the counter ions

Reads a prepared solute structure + topology (e.g. from prep) and builds a hydrated shell with optional neutralising counterions.

Uniform shell — one thickness applied to every non-solvent residue:

parch shellsetup -f solute.gro -p solute.top -o W_init.gro -separateshell no -dshell 4.15 -netcharge 0

Multiple shell regions — any number of (thickness residue-range) groups from a shell-definition file:

parch shellsetup \
    -f solute.gro -p solute.top -o W_init.gro \
    -separateshell yes -shelldef shelldef_setup.txt -netcharge 0

shelldef_setup.txt (one group per line; # comments and blank lines ignored):

# shell_thickness(Å)  residue_range
4.15  1:195
4.50  196:200
4.60  201:220

Shell thickness for setup: For protein: 4.15 Å. For DNA and RNA: 4.50 Å

Validation (separateshell): the file must exist, ranges must be valid, must not overlap, every referenced residue ID must exist in the structure, and all thicknesses must be positive — otherwise the run aborts with a clear message.

Inputs are read from the current directory; all generated files are written to a shell/ subdirectory (the input directory stays clean). The key outputs are the hydrated structure (-o), W_topol.top, and W_ind.ndx, which are also copied into independent run directories shell/mid_1 … shell/mid_N.

Option	Meaning
-f / -p / -o	input structure / input topology / output structure name
-dshell	uniform shell thickness (Å) when -separateshell no (default 4.15)
-separateshell	no (uniform -dshell) or yes (per-group from -shelldef)
-shelldef	shell-definition file (required with -separateshell yes)
-netcharge	system net charge; sets the number of counterions
-nmids	number of mid_* run dirs to create (integer ≥ 3; default 5)
-water	water model tip3p/tip4p/tip5p (selects the solvate box)
-dsi/-dii/-db	ion–solute / ion–ion / solute–box distances (nm)

Auxiliary files (hydrated-ion .gro, INP make_ndx templates) come from -datadir (default: ../backup_annealing).

---

2b. parch membshell — build the hydration shell for a whole membrane

The membrane counterpart of shellsetup. Use it when the "solute" is an entire lipid bilayer (optionally with an embedded membrane protein). The input should be membrane-only (lipids and/or with protein, no water or ions — e.g. straight from prep) and MUST still carry its equilibrated box vectors, because membshell needs the periodic XY size.

It shares every argument with shellsetup (-f/-p/-o, -dshell, -separateshell/-shelldef, -netcharge, -nmids, -water, -watername, -datadir) and differs only in how the box, the counterions, and the interior water are handled:

# neutral membrane, remove waters buried in the core
parch membshell -f membrane.gro -p membrane.top -o W_init.gro \
    -dshell 4.15 -delewater glycerol.txt -netcharge 0

# charged membrane needing many counterions: stack more ion planes
parch membshell -f membrane.gro -p membrane.top -o W_init.gro \
    -delewater glycerol.txt -netcharge -271 -ionplanes 6

Like shellsetup, it solvates and carves a -dshell shell around the lipids, writes everything to a shell/ subdirectory, and copies the key outputs (-o, W_topol.top, W_ind.ndx) into shell/mid_1 … mid_N. Solvation is done in the input's own equilibrated box, so only a physical amount of water is added before the shell is carved.

Non-cubic box. Boxes for membrane setup are elongated along the normal (Z) and keep their periodic lateral (XY) size, so the box is set independently:

• -dbxy pads XY: box_xy = membrane_xy + 2*dbxy. Use 0 (default) to keep the input's periodic XY dimensions exactly.
• -dbz is the gap from the outermost ion plane to the box wall in Z. The full height follows from the ion-plane layout: box_z = 2 * (dbz + (ionplanes/2 - 1)*dii + dsi), with the system centred. This is measured from the system centre, so an embedded protein that protrudes past the bilayer faces does not inflate the box.

Planar counterions. Instead of scattering ions on a sphere, membshell places them in flat layers above and below the membrane:

• -dsi is the distance from the system's Z centre to the innermost ion plane (must exceed the membrane/protein half-thickness, or the ions would land inside it).
• -ionplanes (even: 2, 4, 6, 8; default 2) sets how many layers, split evenly above/below. Each side stacks ionplanes/2 planes outward from centre ± dsi, spaced -dii apart in Z. Ions within a plane are laid on a hexagonal lattice opened up to spread them over the whole footprint, always at least -dii apart; ions in different planes are -dii apart by construction.
• -dii is the minimum in-plane ion–ion distance, and also the plane-plane distance (nm; default 3.0).

If a plane cannot hold all the counterions at the requested -dii, membshell fails immediately (before solvating) with the exact -ionplanes value that would fit — e.g. a heavily charged membrane needing 271 ions cannot fit them in 2 planes at -dii 3.0 and is told to use -ionplanes 6.

Removing water from the membrane core (-delewater). For lipid membrane, after solvate and cut-off the hydration shell, there are usually water molecules placed inside the membrane hydrophobic core. Such waters will be hard to move out druing annealing, so they should be deleted.

-delewater takes a file of leaflet reference atoms whose mean Z marks the bilayer mid-plane; the mean Z of the atoms above / below it defines the upper and lower head-group planes, and every water between those two planes is deleted. The file is either a bare atom-name list or a full MDAnalysis selection:

# glycerol.txt — a full selection (used verbatim when it contains selection keywords)
resname POPC POPE POPS DOPA PIPI and name C1 C2 C3

# a bare atom-name list is also accepted (becomes "name P")
P

Use a full resname ... and name ... selection when the reference atom names are ambiguous across residues — e.g. glycerol carbons C1 C2 C3 also exist (as ring carbons) in cholesterol and glycolipids, so restrict the selection to the glycerophospholipids. Omit -delewater to keep all shell water.

Option	Meaning
-f / -p / -o	input membrane structure / topology / output structure name
-dshell	uniform shell thickness (Å) when -separateshell no (default 4.15)
-separateshell / -shelldef	uniform vs per-group shell thickness (as in shellsetup)
-netcharge	system net charge; sets the number of counterions
-ionplanes	number of ion planes, even 2/4/6/8 (default 2)
-dsi	Z-centre → innermost ion plane distance (nm; default 4.0)
-dii	minimum in-plane ion–ion distance (nm; default 3.0)
-dbxy	XY padding: box_xy = membrane_xy + 2*dbxy (nm; default 0.0)
-dbz	outermost ion plane → box wall gap in Z (nm; default 2.0)
-delewater	leaflet-reference file; deletes waters inside the membrane core
-nmids	number of mid_* run dirs (integer ≥ 3; default 5)
-water / -watername	water model / [ molecules ] water name
-datadir	auxiliary files (hydrated-ion .gro, INP templates)

Note. -delewater uses only the Z coordinate, so for a membrane protein with a water-filled pore it will also delete the pore/channel water. Leave -delewater off (or restore the pore water afterwards) for channel systems.

---

3. parch submit — stage inputs and launch the annealing

For each mid_* run, copies in the simulation inputs (energy-min + annealing .mdp, and the SLURM job script for the chosen partition) and — by default — submits the job with sbatch.

# cpu partition, all mid_*, submit immediately
parch submit -path shell -partition cpu

# gpu partition, first 3 runs, stage only (don't submit)
parch submit -path shell -nmids 3 -partition gpu -launch no

Option	Meaning
-path	the shell/ directory holding the mid_* runs
-nmids	submit mid_1..mid_N; if omitted, all mid_* under -path
-launch	yes (default) submit with sbatch; no stage only
-partition	node partition for the jobs, cpu (default) → evp_<i>_zest3.sh; gpu → evp_<i>_gpu.sh

Each mid_i uses its own annealing protocol (heat_nvt_<i>.mdp) and job script (evp_<i>_zest3.sh / evp_<i>_gpu.sh). **Edit the #SBATCH lines in those mid_*/evp_<i>_*.sh or under bundled backup_annealing/ to change your job requests ** (partition, nodes, walltime, …). Runs that are already finished (w_h.gro present) or not yet set up (W_init.gro missing) are skipped. The protocol/job templates ship for indices 1–5, so this supports up to 5 mid_* runs.

Each run produces em.tpr/.gro and w_h.tpr/.xtc, which the analysis step reads.

---

4. parch analysis — hydration over the annealing ramp

For each mid_* run, counts how many solvent molecules sit within a cutoff of each solute residue, in the energy-minimised structure and across the heating trajectory, and writes b-factor-coloured PDBs at 11 temperature points. Outputs go to mid_*/analysis/.

# uniform cutoff
parch analysis -path shell -da 3.15 -separateshell no

# per-residue cutoffs from a shell-definition file
parch analysis -path shell -separateshell yes -shelldef shelldef_analysis.txt

IMPORTANT NOTICE!!!! -shelldef shelldef_analysis.txt

The shell cutoff for analysis can be different than setup. But the format is consistent with the one used for setup.

Shell cutoff for analysis: For protein: 3.15 Å For DNA and RNA: 4.50 Å

-blocks blocks.txt file — split chosen residues into named atom sub-units (one or more lines per residue range; whitespace-separated, no commas):

# resid_range  block_name  atom1 atom2 ...
1:24   PSU   O5' H5T C5' H5'1 ... P O1P O2P
1:24   NBP   N9 C4 N2 N3 ... N6 H61 H62 H41 H42

Each named residue then yields one result per block (5_PSU, 5_NBP, …); residues not named fall back to whole-residue analysis.

Option	Meaning
-path	the shell/ directory holding the mid_* runs
-da	uniform cutoff (Å) when -separateshell no (default 3.15)
-separateshell	no (uniform -da) or yes (per-residue from -shelldef)
-shelldef	shell-definition file (required with -separateshell yes)
-newwater	extra solvent residue name(s) beyond SOL OPC TP3 T4D T4E
-blocks	block-mapping file to split residues into sub-units (e.g. DNA sugar/base)
-overwrite	re-run even if a mid_*/analysis/ already looks complete

---

5. parch calpv — PARCH values, averaged over runs

Reads the num_ww_temp_<tag>.txt from each mid_*/analysis/, computes the PARCH value per analysis unit (time-correlation integral of the dehydration series scaled by hard_ref), and averages over the runs.

parch calpv -path shell -water_ff charmm_tip3p -nmids 5 -mean_without_min_max yes

Option	Meaning
-path	the shell/ directory with analysed mid_* runs
-water_ff	selects hard_ref (default charmm_tip3p)
-newhardref	user hard_ref value, overrides -water_ff
-nmids	number of runs to include (integer ≥ 3; default 5)
-mean_without_min_max	yes (default) trims lowest+highest run per unit; auto-disabled when < 5 runs
-blocks	block-mapping file (only needed to colour the output PDBs per sub-unit)

hard_ref values. The default hard_ref for each water model/force field is paired with the default parch analysis (the sample-then-sort trend) and is the value used by default:

-water_ff	hard_ref (default)
charmm_tip3p	828602.27
charmm_tip4p	834204.55
charmm_tip4pew	793704.55
charmm_tip5p	886392.05
amber_opc	548924.24
amber_tip3p	813643.94

For any other costumized reference, pass -newhardref <value> (overrides -water_ff). Or add them as new ones in the calpv.py.

Outputs under shell/ (the two averaged flavours are mutually exclusive):

-mean_without_min_max	files written
no	ave_pv_<tag>.txt, std_pv_<tag>.txt, ave_correlation_pv.pdb
yes	ave_excl_pv_<tag>.txt, std_excl_pv_<tag>.txt, ave_excl_correlation_pv.pdb
(always)	parch_summary_<tag>.txt + per-run mid_*/analysis/pv_<tag>.txt, correlation_pv.pdb

parch_summary_<tag>.txt is a per-unit table: unit name mid_1 … mid_N ave std — the residue id, the residue name (GLN, or DG_PSU with blocks), the PARCH value from each run, and the across-run average and std.

---

Auxiliary commands

These two are variants of the pipeline steps above; most users never need them.

parch analysis_na — sort-then-sample hydration trend (DNA/RNA reproduction only)

Drop-in alternative to parch analysis with the same options (-path, -da, -separateshell, -shelldef, -newwater, -blocks, -overwrite) and the same outputs. The only difference is how the monotonic dehydration trend (num_ww_temp) is built:

• parch analysis (sample-then-sort): read the 11 temperature frames, then enforce the non-increasing envelope among only those 11 points. This matches the original PTM/protein scripts (3_all_water_analysis_PTM.py) — the default.
• parch analysis_na (sort-then-sample): enforce the non-increasing envelope over every frame first, then sample the 11 points. Matches sorting_1st_array3D in the old DNA/protein analysis scripts.

parch analysis_na -path shell -da 3.15 -separateshell no

Use it only to reproduce the previous DNA/RNA PARCH work; parch analysis is preferred for all new work. The previous DNA/RNA work share the same hard_ref as protein's work, as listed in previous section. When you feed an analysis_na trend into parch calpv, pair it with correct hard_ref.

parch cval — raw correlation values (useful for getting a new hard_ref)

It reports each unit's time-correlation integral AS IS, without dividing by a hard_ref — i.e. the value parch calpv computes right before it scales by hard_ref to produce the PARCH value. Same inputs as parch calpv (the num_ww_temp_<tag>.txt in each mid_*/analysis/). As there is no normalisation, there is no -water_ff/-newhardref option.

parch cval -path shell -nmids 5 -mean_without_min_max yes

Option	Meaning
-path	the shell/ directory with analysed mid_* runs
-nmids	number of runs to include (integer ≥ 3; default 5)
-mean_without_min_max	yes (default) trims lowest+highest run per unit; auto-disabled when < 5 runs

Outputs under shell/ mirror calpv with a cval prefix (the two averaged flavours are mutually exclusive):

-mean_without_min_max	files written
no	ave_cval_<tag>.txt, std_cval_<tag>.txt
yes	ave_excl_cval_<tag>.txt, std_excl_cval_<tag>.txt
(always)	cval_summary_<tag>.txt + per-run mid_*/analysis/cval_<tag>.txt

---

Notes

• Run any command with -h for the full option list.
• prep/analysis/analysis_na/calpv/cval need only MDAnalysis; shellsetup and membshell also need gmx; submit needs SLURM (sbatch).
• The -shelldef format is identical across shellsetup, analysis and (for prep) the residue ranges it prints, so the same file/ranges carry through.