# HiContacts
<!-- badges: start -->
[](https://lifecycle.r-lib.org/articles/stages.html#experimental)
[](https://github.com/js2264/HiContacts/actions/workflows/check-bioc.yml)
[](https://github.com/js2264/HiContacts/actions)
<!-- badges: end -->
HiContacts provides tools to import `(m)cool` matrices in R and work with them there.
Rather than creating redundant classes, it relies on pre-existing Bioconductor objects, namely `InteractionSet`, `GenomicInterations` and `ContactMatrix` (`Lun, Perry & Ing-Simmons, F1000Research 2016`).
## Installation
```r
remotes::install_github('js2264/HiContacts')
```
## Import a .(m)cool file as `contacts`
`contacts()` can be used to import a Hi-C matrix.
```r
cool_file <- system.file("extdata", 'CH112.cool', package = 'HiContacts')
range <- 'I:20000-80000' # range of interest
contacts <- contacts(cool_file, focus = range)
contacts
```
`cool2gi` works with `.mcool` files as well: in this case, the user needs to specify the resolution at which count values are recovered.
```r
mcool_file <- system.file("extdata", 'CH112.mcool', package = 'HiContacts')
range <- 'I:20000-80000' # range of interest
lsCoolResolutions(mcool_file)
contacts <- contacts(mcool_file, focus = range) # This throws an error!
contacts <- contacts(mcool_file, focus = range, res = 1000)
contacts
```
## Plotting matrices
### Plot matrix heatmaps
> Diagonal style
```r
plotMatrix(contacts, use.assay = 'raw')
plotMatrix(contacts, use.assay = 'balanced', limits = c(-4, -1))
```
> Plot matrix of correlation
```r
contacts(mcool_file, focus = 'II', res = 1000) %>%
autocorrelate() %>%
plotMatrix(scale = 'linear', limits = c(-1, 1))
```
> Plot signal over expected
```r
contacts(mcool_file, focus = 'II', res = 1000) %>%
detrend() %>%
plotMatrix(use.assay = 'expected')
contacts(mcool_file, focus = 'II', res = 1000) %>%
detrend() %>%
plotMatrix(use.assay = 'detrended', scale = 'linear', limits = c(-2, 2))
```
### Plot loops
```r
loops <- system.file("extdata", 'CH112_1kb.bedpe', package = 'HiContacts') %>%
rtracklayer::import() %>%
InteractionSet::makeGInteractionsFromGRangesPairs()
contacts(mcool_file, focus = 'IV', res = 1000) %>%
detrend() %>%
plotMatrix(use.assay = 'detrended', loops = loops, scale = 'linear', limits = c(-2, 2))
```
### Plot borders
```r
borders <- rtracklayer::import('AT507_borders.bed')
contacts(mcool_file, focus = 'IV', res = 1000) %>%
detrend() %>%
plotMatrix(use.assay = 'detrended', loops = loops, borders = borders, scale = 'linear', limits = c(-2, 2))
```
### Many plots at once
```r
p <- cowplot::plot_grid(
contacts(mcool_file, focus = 'IV', res = 1000) %>% plotMatrix(use.assay = 'raw'),
contacts(mcool_file, focus = 'IV', res = 1000) %>% plotMatrix(use.assay = 'balanced'),
contacts(mcool_file, focus = 'IV', res = 1000) %>% detrend() %>% plotMatrix(use.assay = 'expected', scale = 'linear'),
contacts(mcool_file, focus = 'IV', res = 1000) %>% detrend() %>% plotMatrix(use.assay = 'detrended', scale = 'linear', limits = c(-3, 3)),
contacts(mcool_file, focus = 'IV', res = 1000) %>% detrend() %>% plotMatrix(use.assay = 'detrended', scale = 'linear', limits = c(-3, 3), loops = loops),
contacts(mcool_file, focus = 'IV', res = 8000) %>% autocorrelate() %>% plotMatrix(scale = 'linear', limits = c(-1, 1))
)
ggsave('tmp.pdf', width = 20, height = 10)
```
## Analysis in R
### Auto-correlation
```{r}
```
### Ratio of maps
```{r}
divide() %>% smoothen()
```
### Virtual 4C
```r
contacts <- contacts(mcool_file, res = 1000)
v4C <- virtual4C(contacts, viewpoint = GRanges('V:150000-170000'))
gg4C(v4C, aes = aes(x = center, y = score, col = chr)) +
facet_wrap(~chr, scales = 'free_x')
```
### Cis-trans ratios
```r
contacts <- contacts(mcool_file, res = 1000)
cis_trans(contacts)
```
### Compute P(s)
```r
pairs <- system.file("extdata", 'CH112.pairs', package = 'HiContacts') %>%
pairs2gi()
ps <- getPs(pairs)
ggPs(ps, aes(x = binned_distance, y = norm_p), xlim = c(5000, 5e5), ylim = c(1e-10, 1e-4))
```
### Compute scalograms
```{r}
pairs <- pairs2gi('tmp2.gz')
df <- getScalogram(pairs, ylim = c(1e4, 1e7))
p <- ggScalogram(df, aes(x = binned_pos, ymin = y0, ymax = dist_quantile, group = prob, fill = prob), ylim = c(1e4, 1e7))
```
### APA : Aggregated Plot Analysis
```{r}
```
