---
title: "celltoprotein -- connecting Cell and Protein Ontologies"
author: "Vincent J. Carey, stvjc at channing.harvard.edu"
date: "`r format(Sys.time(), '%B %d, %Y')`"
vignette: >
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteIndexEntry{celltoprotein -- connecting Cell and Protein Ontologies}
  %\VignetteEncoding{UTF-8}
output:
  BiocStyle::html_document:
    highlight: pygments
    number_sections: yes
    theme: united
    toc: yes
---

# Introduction

In a pair of papers from the Ventner Institute, 
[Bakken et al.](https://link.springer.com/article/10.1186/s12859-017-1977-1) and
[Aevermann et al.](https://academic.oup.com/hmg/article/27/R1/R40/4953379) discuss
ontological implications of single-cell transcriptomics.  A process of cell type
definition via "necessary and
sufficient marker gene" enumeration is introduced.

In this vignette we indicate how Cell Ontology, Relational Ontology,
and Protein Ontology can be connected to assess formal relationships
between declared cell types and plasma membrane features that can
play a role in cell type definition.

# Given a cell type, what proteins are noted as parts of its plasma membrane?

Connect to the relational ontology and search for CURIEs related
to "plasma membrane".
```{r lkon1,message=FALSE}
library(ontoProc2)
library(DT)
ro <- semsql_connect(ontology = "ro")
search_labels(ro, "plasma membrane")
```

We have a helper resource for finding exact Cell Ontology names
of cell types.

```{r lkct1}
data("tag2cn", package = "ontoProc2")
cd8reg <- grep("CD8-positive.*regulatory", tag2cn, value = TRUE)
cd8reg
```

Now with these cell type identifiers, we can search for the
proteins identified as "part of plasma membrane".
We need to use the CURIEs for precision.  THIS IS BLOCKED
UNTIL WE HAVE A SUBSET OF PR DATA TO ILLUSTRATE AS THE
PR DOWNLOADS ARE TOO SLOW.
```{r lkct2, message=FALSE,eval=FALSE}
prtab <- get_present_pmp(names(cd8reg))
datatable(prtab)
```

# Given a protein, what cell types are asserted to possess it as a membrane part?

We pick two proteins and look for associated cell types.  BLOCKED AS ABOVE.

```{r lkpr1, message=FALSE,eval=FALSE}
prs <- c("PR:000001094", "PR:000001380")
clk <- try(cells_with_pmp(prs))
if (inherits(clk, "try-error")) message("it is necessary to allow a large download of Protein Ontology for this chunk to run") else datatable(clk)
```

# Some details

The "entailed edge" table of the Semantic SQL representation of
Cell Ontology includes all assertions that are derivable from base
axioms of the ontology.  

```{r lkentedg, message=FALSE}
cl <- semsql_connect(ontology = "cl")
cl
library(dplyr)
tbl(cl@con, "entailed_edge")
tbl(cl@con, "entailed_edge") |> count()
```

We can look for statements that have "RO:0002104" as predicate:
```{r domo}
tbl(cl@con, "entailed_edge") |>
  filter(predicate == "RO:0002104") |>
  as.data.frame() |>
  filter(grepl("PR:", object)) |>
  arrange(subject) |>
  datatable()
```

Disconnect databases.

```{r dodisc}
disconnect(cl)
disconnect(ro)
```



# Session information

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