insertIon method about compounddb HOT 4 CLOSED

What I meant with CompDb is missing is the following:

idb <- IonDb(dbcon = icon, CompDb = cdb)

is supposed to create a new IonDb database from content taken from the CompDb cdb. We should however also allow a call like:

idb <- IonDb(dbcin = icon)

in which case icon is expected to be the database connection to an already existing IonDb (with all the required tables). This call should allow users to load a previously created database.

So, the IonDb function should be something like:

IonDb <- function(dbcon, CompDb) {
    if (!missing(CompDb)) {
        ## Call a function that copies the content from CompDb over to dbcon
        ## and create the other required tables for IonDb in dbcon
    }
    ## assign dbcon to @dbcon, check validity of the database and return it
}

from compounddb.

I try to explain the expected workflow: the user is supposed to have created (or downloaded) first a CompDb with compound annotations from e.g. HMDB. The goal would be to add also ion information to that database an hence convert the database to a IonDb.

One possible workflow would be, given that cdb is an existing CompDb and ions is a data.frame with ion information the user wants to add:

## create a new SQLite database that will contain the IonDb
idbcon <- dbConnect(SQLite(), "my_ion_db.sqlite")
## add ions to the database:
idb <- insertIons(cdb, ions, dbcon = idbcon)
## What this call needs to do:
## - copy all content from the original `cdb` to the new database
## - create the ms_ion table and add the ions to that
## - return an IonDb object

If the user wants to add additional ions it would be much easier, because he/she has already an IonDb:

## no need here to specify a different database or copy content, because `idb` does already have
## the required table and is expected to be read-write
idb <- insertIons(idb, ions)

I agree that this workflow is not that intuitive - because we're changing from a CompDb to a IonDb with the insertIons call. Maybe we should think of an alternative @andreavicini ?

from compounddb.

What about: insertIons is only defined for IonDb objects? The user would need to create first a IonDb database/object with the IonDb constructor function:

icon <- dbConnect(SQLite(), "my_ion_db.sqlite")
idb <- IonDb(dbcon = icon, CompDb = cdb)

A call like the one above is supposed to copy all tables from the CompDb over to dbcon and create in addition the ms_ion table and return the IonDb object with the connection to the new database (dbcon). If CompDb is missing the specified database connection is supposed to be for a database containing already the tables needed for an IonDb.

From here on insertIons could be called to insert ions in the new database.

Would that be a better workflow? @andreavicini what do you think?

from compounddb.

I like this second approach and maybe it's a bit clearer. I'm not sure if I understood what you mean in the case CompDb is missing. Should it have like empty tables for each of the required tables?

from compounddb.