---
title: "crisprScoreData"
author: 
- name: Jean-Philippe Fortin
  affiliation: Data Science and Statistical Computing, gRED, Genentech
  email: fortin946@gmail.com
date: "`r Sys.Date()`"
output: 
  BiocStyle::html_document:
    toc_float: true
    number_sections: true
vignette: >
  %\VignetteIndexEntry{crisprScoreData}
  %\VignetteEngine{knitr::rmarkdown}
  \usepackage[utf8]{inputenc}
---


# Installation from Bioconductor

`crisprScoreData` can be installed from the Bioconductor devel
branch using the following commands in a fresh R session:

```{r, eval=FALSE}
if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install(version="devel")
BiocManager::install("crisprScoreData")
```


# Exploring the different data in crisprScoreData

We first load the `crisprScoreData` package:

```{r}
library(crisprScoreData)
```

This package contains several pre-trained models for different
on-target activity prediction algorithms to be used 
in the package *crisprScore*.

We can access the file paths of the different 
pre-trained models directly with named functions:

```{r}
# For DeepHF model:
DeepWt.hdf5()
DeepWt_T7.hdf5()
DeepWt_U6.hdf5()
esp_rnn_model.hdf5()
hf_rnn_model.hdf5()

# For Lindel model:
Model_weights.pkl()
```


Or we can access them using the *ExperimentHub* interface:

```{r}
eh <- ExperimentHub()
query(eh, "crisprScoreData")
eh[["EH6127"]]
```

For details on the source of these files, and on their construction
see `?crisprScoreData` and the scripts:

* `inst/scripts/make-metadata.R`
* `inst/scripts/make-data.Rmd`

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