Table 3.

Results from ten-temperature melting analysis of ${\text{hPTH}}_1\text{R}$ variants.

Module	Variant	$T_m$ (${}^{\circ }$C)${}^{\mathrm{a}}$	$T_m$ error (${\pm }^{\circ }$C)${}^{\mathrm{b}}$	$\mathrm{\Delta }{T}_m$ (${}^{\circ }$C)	$\mathrm{\Delta }{T}_m$ error (${\pm }^{\circ }$C)${}^{\mathrm{c}}$	Repeats (n)	Status
n/a	Wild-type	38.7	0.8	0	1.1	3	n/a
Deep-sequence	G188Y	41.2	0.7	2.5	0.7	3	Stabilising
Deep-sequence	M189L	39.1	0.9	0.4	0.9	3	Neutral
Deep-sequence	G188K	40.3	1.2	1.6	1.2	3	Stabilising
Deep-sequence	A274D	39.1	0.1	0.4	0.1	3	Neutral
Model-based	S198M	42.1	0.8	3.4	1.1	3	Stabilising
Model-based	F291T	40.9	0.9	2.2	1.2	3	Stabilising
Model-based	D251R	41.7	1.5	3.0	1.6	3	Stabilising
Model-based	E260R	39.7	1.5	1.0	1.7	3	Neutral
Data-driven	Q401A	40.7	0.2	2.0	0.2	3	Stabilising
Data-driven	F288A	41.5	0.7	2.8	0.7	3	Stabilising
Data-driven	E391A	39.4	0.2	0.7	0.2	3	Neutral
Data-driven	G323A	41.5	0.6	2.8	0.6	2	Stabilising

${}^{\mathrm{a}}$Average $T_m$ was calculated from individual $T_m$ estimated for each individual repeat by fitting with a four-parameter dose-response curve (variable slope) by non-linear least-squares fitting in the python package scipy.stats.

${}^{\mathrm{b}}$For n > 1: standard error of the mean (SEM) shown.

${}^{\mathrm{c}}$For n > 1: error calculated and propagated as detailed in “Methods and materials” section.