README.txt

Data repository for study
Elbau et al. The hemodynamic response function rapidly changes under psychosocial
stress in association with genetic and endocrine stress response markers. PNAS,
accepted 08/2018

The repository contains the following 8 file types:

[1] First level models of the original (N=59,
comprising 39 experimental and 20 control subjects) and the replication
sample (N=39) (only experimental runs).

[2] Two key scripts that allow to calculate (a) the HRF-PL values
according to Henson et al. 2002 and (b) lag-corrected amplitude
values according to Calhoun et al. 2003. Note that prior to 
statistical analyses these files were spatially smoothed.

[3] Two scripts destination.m and hrf_generation.m that make
the generation of the sigmoid transformation function transparent.

The script hrf_generation.m compares different beta2/beta1 ratios
against the empirically caused shift of the HRF peak when both
the canonical and the (orthogonalized) first derivative are
added. The resulting function may be newly produced and stored,
yet is also directly provided as file [4] = destination.mat.
Note that the script relies on the function spm_hrf.m 
(in the version of SPM8).

The script destination.m optimizes C and D within certain
ranges, after it was noted that the original values of C =1.78
and D = 3.10 (Henson et al. 2002) showed deviations from the
goldstandard. The error was already minimal by values C = -2 and
D = -0.9, yet the plot was then systematically shifted, particularly
in the y-direction. Thus, two correction coefficients were 
introduced to correct for this shift. The eventually used
values were: C = -2, D = -0.7, Xcorr = 0.25, Ycorr = -0.5.

[5] The file destination.mat as explained in [3].

[6] All MNI space transformed and discretized anatomical parcellations
(FreeSurfer, see methods) to allow for regional value extraction.
Files are labelled 'rwfsl_atlas_clean.img'. They are in the folders 'T1WI' (original sample) or 'mri' respectively.

[7] The background image used for result depiction, generated from
a mean image of all DARTEL normalized T1 images, combined with a
transparent overlay. The latter was defined (as specified in the
paper) from a combination of three F contrasts on the lag-corrected
amplitude effects.

[8] Cortisol data per subject (original sample)

Questions on this material may be directed to
immanuel_elbau@psych.mpg.de or saemann@psych.mpg.de

30th August 2018