mashup.R

---
title: "Searching Nanopubs"
output: html_notebook
---

The first step is to set up a search for the NanoPublications
REST API. We use the `curl` library for that, and the `jsonlite`
to process the JSON.

```{r}
library(curl)
library(jsonlite)
```

We also define where the REST API id found:

```{r}
searchAPI = "http://grlc.io/api/peta-pico/nanopub-api/find_latest_nanopubs_with_uri"
```

Let's say we are searching for ENSG00000181852:

```{r}
gene = "http://identifiers.org/ensembl/ENSG00000181852"
```

We define a handle with the information to make the API call. We want the results
returned as JSON and we use the `ref` parameter to pass the gene IRI:

```{r}
npHandle <- new_handle()
handle_setopt(npHandle, customRequest="GET")
handle_setheaders(npHandle,
  "Accept" = "application/json"
)
```

With this search call defined, we can run it against the server, and retrieve
our JSON:

```{r}
r <- curl_fetch_memory(
  paste(searchAPI, "?ref=", curl_escape(gene), sep=""), npHandle
)
jsonReply = rawToChar(r$content)
data = fromJSON(jsonReply)
```

We can count the number of returned nanopublications:

```{r}
nanopubs = data$results$bindings
if (length(nanopubs) == 0) nanopubs = as.vector(nanopubs)
if (length(nanopubs) == 1) nanopubs = as.vector(nanopubs[[1]][[2]])
cat(length(nanopubs))
```

But we cannot know IRI scheme or even from which database gene identifiers are used in
nanopublications. Therefore, we must use multiple IRIs for the same gene, and the BridgeDb
Identifier Mapping Service (IMS) gives those to us:

```{r}
mapUriService = "http://localhost:8081/QueryExpander/mapUri"

h <- new_handle()
handle_setopt(h, customrequest = "POST")
handle_setform(h, Uri=gene, format="application/json")
r <- curl_fetch_memory(mapUriService, h)
jsonReply = rawToChar(r$content)
data = fromJSON(jsonReply)
mappings = length(data$Mapping$targetUri)
cat(paste("Mappings found:", mappings, "\n"))
data$Mapping$targetUri
```

For convenience, we define a function that will iterate over a list of IRIs to get nanopublications,
based on the earlier used code:

```{r}
getNanoPubs <- function(seedIRI) {
  collectedNanopubs = c()
  
  # stuff for the identifier mapping
  mapUriService = "http://localhost:8081/QueryExpander/mapUri"
  h <- new_handle()
  handle_setopt(h, customrequest = "POST")
  handle_setform(h, Uri=seedIRI, format="application/json")
  
  # stuff for the NP searching
  searchAPI = "http://grlc.io/api/peta-pico/nanopub-api/find_latest_nanopubs_with_uri"
  npHandle <- new_handle()
  handle_setopt(npHandle, customRequest="GET")
  handle_setheaders(npHandle,
    "Accept" = "application/json"
  )
  
  # first, look up equivalent IRIs
  r <- curl_fetch_memory(mapUriService, h)
  jsonReply = rawToChar(r$content)
  data = fromJSON(jsonReply)
  mappings = length(data$Mapping$targetUri)
  cat(paste("Mappings found:", mappings, "\n"))
  iris <- data$Mapping$targetUri
  
  # next, look up each mapped IRI
  for (iri in iris) {
    cat(paste("Processing:", iri, "... "))
    r <- curl_fetch_memory(
      paste(searchAPI, "?ref=", curl_escape(iri), sep=""), npHandle
    )
    jsonReply = rawToChar(r$content)
    data = fromJSON(jsonReply)
    nanopubs = data$results$bindings
    if (length(nanopubs) == 0) nanopubs = as.vector(nanopubs)
    if (length(nanopubs) == 1) nanopubs = as.vector(nanopubs[[1]][[2]])
    if (length(nanopubs) > 0) {
      cat(paste(" nanopubs:", length(nanopubs)))
      collectedNanopubs = c(collectedNanopubs, nanopubs)
    }
    cat("\n")
  }
  return(collectedNanopubs)
}
```

Test:

```{r}
getNanoPubs("http://identifiers.org/entrez.gene/672")