minimal.Rmd

---
title: "Minimal examples for `solarius` R package"
author: "Andrey Ziyatdinov"
date: "`r Sys.Date()`"
output:
  html_document:
    theme: united
    toc: true
    toc_depth: 3
    number_sections: true
  pdf_document:
    toc: true
    toc_depth: 3
    number_sections: true
---

```{r chunk_opt, echo = FALSE}
#opts_chunk$set(eval = TRUE)
#opts_chunk$set(fig.path = "figures-minimal/")
```

# `dat30` data set 

```{r, eval = FALSE}
help(dat30, package = "solarius")
```

# All code in one block

```{r min_example, cache = TRUE}
# load library
library(solarius)

# load data set
data(dat30)

# univariate polygenic model
mod1 <- solarPolygenic(trait1 ~ 1, dat30)

# bivariate polygenic model
mod2 <- solarPolygenic(trait1 + trait2 ~ 1, dat30,
  polygenic.options = '-testrhoe -testrhog')

# specify directory with IBD matrices and run linkage model
mibddir <- system.file('extdata', 'solarOutput',
  'solarMibdsCsv', package = 'solarius') 
link <- solarMultipoint(trait1 ~ 1, dat30,
  mibddir = mibddir, chr = 5)

# run association model in parallel
assoc <- solarAssoc(trait1 ~ 1, dat30, cores = 2,
  snpcovdata = genocovdat30, snpmap = mapdat30)
```  

# Results

Estimation of the heritability of `trait1` (`mod1` model) is `r sprintf("%.2f", with(mod1$vcf, Var[varcomp == "h2r"]))`
$\pm$ `r sprintf("%.2f", with(mod1$vcf, SE[varcomp == "h2r"]))`
with p-value `r sprintf("%.2e", with(mod1$vcf, pval[varcomp == "h2r"]))`.

Estimation of the genetic correlation between `trait1` and `trait2` (`mod2` model) 
is `r sprintf("%.2f", with(mod2$vcf, Estimate[varcomp == "rhog"]))`
$\pm$ `r sprintf("%.2f", with(mod2$vcf, SE[varcomp == "rhog"]))`
with p-value `r sprintf("%.2e", with(mod2$lf, pval[model == "rhog0"]))`.

Estimation of the environmental correlation between `trait1` and `trait2` (`mod2` model) 
is `r sprintf("%.2f", with(mod2$vcf, Estimate[varcomp == "rhoe"]))`
$\pm$ `r sprintf("%.2f", with(mod2$vcf, SE[varcomp == "rhoe"]))`
with p-value `r sprintf("%.2f", with(mod2$lf, pval[model == "rhoe0"]))`.

The highest LOD score in the linkage analysis of `trait1` on Chromosome 5 
is `r sprintf("%.2f", max(link$lodf$LOD))` at `r link$lodf$pos[which.max(link$lodf$LOD)]` cM position.

The only significant SNP in the association analysis of `trait1` 
on a set of `r nrow(assoc$snpf)` (synthetic) SNPs is 
`r as.character(assoc$snpf$SNP[which.min(assoc$snpf$pSNP)])`
with p-value `r sprintf("%.2e", min(assoc$snpf$pSNP))`.

# Polygenic model (univariate)

The model with covariates can be fitted in the following way.

```{r mod0, cache = TRUE}
mod0 <- solarPolygenic(trait1 ~ age + sex, dat30, covtest = TRUE)
```

`print` method applied to `mod0` shows that none of the covariates is significant.

```{r print_mod0}
mod0
```

The test statistics and p-values for the covariates are stored in `cf` (**c**covariate **f**rame) slot.

```{r mod0_cf}
mod0$cf
```

The estimations of heritabilities in `mod0` and `mod1` models are slightly different.

```{r mod1}
mod1
```

# Polygenic model (bivariate)

The `print` method applied to `mod2` shows estimation of the main parameters of the model.

```{r mod2}
mod2
```

The same values printed above are extracted and saved into `vcf` (**v**ariance **c**omponent **f**rame) slot .

```{r mod2_vcf}
mod2$vcf
```

The information about the test statistic and p-values are given in `lf` (**l**ikelihood **f**rame) slot.

```{r mod2_lf}
mod2$lf
```

# Linkage model

Table of the results with LOD scores and marker information is stored in `lodf` (**LOD** scores **f**rame) slost.

```{r link}
link$lodf
```

Summary method reports the maximum LOD score.

```{r sum_link}
summary(link)
```

The plot method graphically shows the results.

```{r plot_link}
plot(link)
```

# Asssociation model

`summary` method applies the Bonferroni correction with a given value of `alpha` (the default value is 0.05).

```{r assoc}
summary(assoc, alpha = 0.05)
```

Table of the results is stored in `snpf` (**SNP** **f**rame) slot.

```{r assoc_snpf}
head(assoc$snpf)
```

The default plot method is Manhattan plot.
The black dashed line shows the Bonferroni correction level at a given value of `alpha` (the default value is 0.05).

```{r plot_assoc_man}
plotManh(assoc, alpha = 0.01)
```

This call is equavalent to `plot(assoc, alpha = 0.01)`,
as `plot` method applied to `assoc` object calls `plotManh` function.

The second plot available for `assoc` object is QQ-plot.
The inflation parameter $\lambda$ (median method) is computed and dipicted at the bottom of the figure.

```{r plot_assoc_qq}
plotQQ(assoc)
```

This call is equavalent to `plot(assoc, "qq")`,
as `plot` method (applied to `assoc` object) with the second argument equal to `"qq"` calls `plotQQ` function.

The procedure of SNPs annotation is possible via `rsnps` R package (loaded by `annotate` function invisibly to the user).

```{r annot_assoc}
tab <- annotate(assoc)
tab
```

Annotation of the synthetic SNPs generated for `dat30` data sets returns an empty data frame,
since the names expected by `rsnps` R package must begin with *rs* symbols.

Meantime, annotation of the genome-wide significant SNPs 
from [GWAS of FXIc phenotype in GAIT1 data set](http://ugcd.github.io/solarius/vignettes/modelsGAIT1.html#fxi)
(real data) is accomplished as following.

```{r annot_F11_gait1, cache = TRUE}
snps <- c("rs710446", "rs4241824", "rs4253399")
tab <- annotate(snps)
tab
```

In this case `annotate` function takes a vector of characters as its first argument.

# R session info

```{r session_info}
sessionInfo()
```

# License

This document is licensed under the Creative Commons Attribution 4.0 International Public License. 

[![Creative Commons License](http://i.creativecommons.org/l/by/4.0/88x31.png)](http://creativecommons.org/licenses/by/4.0/)