(used on page 3, defined on page 9)

np.mean should have been np.sum for error accumulation.

In the branch colvarsIssue, stepD input/run.namd

#############################################################

COLVARS

#############################################################

colvars on
if {$useSampleFiles} {
colvarsConfig ../../stepA_create_DBC/sample_outputs/DBC_restraint.colvars
} else {
colvarsConfig ../../stepA_create_DBC/outputs/DBC_restraint.colvars
}
cv configfile ../../common/protein_tilt.colvars

#############################################################

FEP PARAMETERS

#############################################################

source ../../common/fep.tcl

alch on
alchType FEP

results in the error:

TCL: Setting parameter alch to on
FATAL ERROR: Alchemy must be enabled at startup to disable and re-enable in script.
FATAL ERROR: Alchemy must be enabled at startup to disable and re-enable in script.
[Partition 0][Node 0] End of program

BUT commenting out "cv configfile ..." runs normally (but without the Colvars defined in protein_tilt.colvars.

We need to clarify that these lines are essential:

colvars on
colvarsConfig $CVconfig

But this line can be removed if the user doesn't have protein restraints:

cv configfile ../../common/protein_tilt.colvars

1. Turn off auto-update selections
2. Remind users to keep it off in the text

• Fix kJ/mol instead of kcal/mol in notebook
• Step A.1.d : say where the « atomPermutation » line should be added
• Step B.2.a and other namd runs: change path of log file, add error output
• Step C.3.a : Error in the Colvar module, due to typo in TI.tcl « outputEnergies on » instead of « outputEnergy on »
• Step E.6 : The abscissa axis is in mM in the PDF tutorial but the notebook outputs it in μM

outputEnergies should be outputenergy

The notebook currently exposes a lot of nitty gritty code to the user which should be encapsulated and (eventually) moved to the safep package.

Some variable names are reused in the notebook. Those should be changed to avoid mismatched analysis (especially if the user isn't running everything sequentially).

• broken links
• "constant constant"
• "TI" is used before "thermodynamic integration"
• Sign error in caption of thermodynamic cycle
• Axis labels for COM and DBC time course
• Missing resids for system building (truncation of lysozyme)
• Cite Zacharias .. McCammon 94 and Ebrahimi & Henin 22 in the discussion of softcore potential
• Clarify KDE in the summary figure
• Remove x from the dG_V function and just use $L~V_R$
• Y axis label of bottom plot in figure 7 (should be partial G/partial lambda)
• emphasize reversibility in discussion of IDWS (justification, appendix B1)

I'm already starting on this in refactorTI here and alternativeTI on the safep repo.

1. The TI blocks in the notebook are long and hard to read
2. The notebook block requires manual editing for alternative lambda schedules, force constants, etc. (error-prone!)
3. The back-end assumes an accurate lambda schedule. The schedule is currently generated in a hacky way in the notebook, but is WRONG for many samples near the points that the lambda value changes. This doesn't affect the overall estimate for large data sets, but shouldn't be present.