simulation exercise and documentation

docs/output files.rst

@@ -213,25 +213,24 @@ along with the parental polygenic scores based on observed and/or imputed parent
 The :ref:`pgs.py <pgs.py>` script will output a PGS file, which is a white-space delimited text
 file with columns: FID (family ID), IID (individual ID), FATHER_ID, MOTHER_ID, proband PGS, paternal PGS, maternal PGS. 
 (If the PGS has been computed from parental genotypes imputed from siblings alone, it will output the imputed parental PGS,
- the sum of paternal and maternal PGS values, instead of the paternal and maternal PGS values separately.)
+the sum of paternal and maternal PGS values, instead of the paternal and maternal PGS values separately.)
 
 Here, the family ID is the same as used internally by *snipar*, so that individuals who share a family
 ID are full-siblings. The FATHER_ID and MOTHER_ID columns are set to NA when the parents are not genotyped. 
 The proband PGS column gives the PGS value for the individual given by the IID in that 
 row. The paternal and maternal PGS columns give the PGS values of the individual's father and mother respectively,
 and these values can be computed from either imputed or observed parental genotypes. 
 
-Note that if the PGS was computed with the '--fit_sib' option, the columns will be 
-FID (family ID), IID (individual ID), proband PGS, sibling PGS, paternal PGS, maternal PGS. The
-sibling PGS column is the PGS computed from the average of the individual's siblings' genotypes. 
+If the PGS was computed with the '--fit_sib' option, the output will include a column
+for the average of the proband's siblings' PGS values.
 
 PGS effects
 -----------
 .. _pgs_effects:
 
 The :ref:`pgs.py <pgs.py>` script can be used to compute the direct and population effects of the PGS
 along with the non-transmitted coefficients (NTCs). By default, the script will fit both a one-generation
-model (regression of phenotype onto proband PGS alone, which estimates the population effect of the PGS) 
+model (regression of phenotype onto proband PGS, which estimates the population effect of the PGS) 
 and a two-generation model (regression of phenotype
 onto proband, paternal, and maternal PGS; or proband and combined parental PGS). 
 

docs/simulation.rst

@@ -92,16 +92,16 @@ This will output a population effect estimate (1 generation model) to direct_v1.
 direct effect and average NTC estimates to (2 generation model) to direct_v2.2.effects.txt. 
 
 Unlike when using the true direct genetic effects as weights, the direct effect of the PGS estimated
-from noisy weights (in direct_v1.1.effects.txt) will be smaller than the population effect (direct_v2.2.effects.txt).
+from noisy weights (in direct_v1.1.effects.txt) will be smaller than the population effect (direct_v1.2.effects.txt).
 This is because the PGS does not capture all of the heritability due to estimation error in the weights. 
 The PGS has its population effect inflated (relative to its
 direct effect) by assortative mating, which induces a correlation of the PGS with the component of the heritability
 not captured by the PGS due to estimation error. This inflation is not captured by the direct effect of the PGS
 because chromosomes segregate independently during meiosis. (In this simulation, all causal SNPs segregate independently.) 
 Here, the ratio between direct and population effects of the PGS should be around 0.87. 
 
-One should also observe a statistically significant average parental NTC (in direct_v2.2.effects.txt) of the PGS from 
+One should also observe a statistically significant average parental NTC (in direct_v1.2.effects.txt) of the PGS from 
 the two-generation model despite the absence of parental indirect genetic effects in this simulation. Here,
 the ratio between the average NTC and the direct effect should be around 0.15. This demonstrates
-that statistically significant average NTC estimates cannot be interpreted as unbiased estimates of 
+that statistically significant average NTC estimates cannot be interpreted as demonstrating
 parental indirect genetic effects, especially for phenotypes affected by assortative mating.