OHDSI Home | Forums | Wiki | Github

Tantalus / DataQualityDashboard implementation issues


(Mark Seal) #1

For the AOU program that we are involved in, we need to have a local instance of Tantalus and DataQualityDashboard running locally. We now have over 40 hours invested in just trying to get the software to run with no success and we have not been able to find anywhere on GitHub, or in this forum, where there are clear instructions on what needs to be done to get these to work.
Where can we get clear instructions on how to build/implement these packages?

There are some caveats:

  • We are on a Windows server
  • We have a very tight security policy that will not allow me to post complete error messages that include any path/build details.
  • There is no-one on the staff that has ever worked in R.
  • We have both the free version of RStudio (which keep complaining about none free packages) and the full developers version of both PyCharm and IntelliJ to work on the R code in.

We would be greatly obliged for any assistance.

My bona Fide, I have over a decade of building enterprise level MDM software in Java/Scala/Python/Postgres and the last 3 years in C#/SQLServer working with medical analytics and reporting.


(Jose Posada) #2

Hi @Mark,

Are you trying to run this on BQ? If yes my advice is to use this docker

https://hub.docker.com/r/odysseusinc/rstudio-ohdsi

Also, the following tips may help if you are on BQ:

  1. Use a service account that only has access to the dataset you want to query and nothing else.
  2. The BQ SQL is not rendering correctly for Tantalus. I ask @Konstantin_Yaroshove to submit a pull request to fix this. The EXCEPT clause is not rendering properly for BQ.

The DQ dashboard runs pretty smoothly and does not require any changes at all from what is inside the docker. Only doing an install_github should do it


(Mark Seal) #3

We do not have a Docker install. This would have to go through several levels of management approval and a security review before it could be done. Whilst I am in agreement that this would be the better way to go, I do not think that this will be allowed.

I am a Java/C family programmer… I am sorry, I do not know what BQ is. When I did a quick google, it leads me to a statistical function, so I am assuming that this is something different.

I first tried running this from cmd line, then from RStudio and then from PyCharm.

We are not even getting to the point of SQL rendering. When I step through the code in PyCharm, I get to this line before it breaks.
connectionDetails <- createConnectionDetails("xxx") with the xxx as local details that I cannot show. Here is the error message I get at this point.
The folder location pathToDriver = ' ' does not exist. Please set the folder to the location containing the JDBC driver. You can download most drivers using thedownloadJdbcDrivers()function.

We are not using JDBC drivers nor does this command work from the R cmd prompt. This may be very simple, but for what ever reason it is escaping me.

Thank you very much.

EDIT: I changed connectionDetails = createConnectionDetails("xxx") to connectionDetails <- createConnectionDetails("xxx") as PyCharm complained that the former was a syntax error. I did this trying to fix the error, so this is not the source of the problem.


(Jose Posada) #4

hi @Mark,

I think all databases need a JDBC driver in order to use the DatabaseConnector package. So I believe you need to download or specify the location of the drivers passing this parameter pathToDriver = to createConnectionDetails. BQ stands for Bigquery. Which database are you using?


(Mark Seal) #5

We are using SQL Server with ODBC drivers and the connection was set up to work off of active directory ( for security reasons).

I will pass this on the the Database Admin as I have no control over this aspect.

Thank you


(Jose Posada) #6

ok That is the issue. OHDSI packages do not work with ODBC drivers only JDBC.


t