DrugBank network mining example

Can you be more specific about the configuration of the XML reader node to read the drugbank.xml file?

I receive a warning:

WARN XML Reader An empty prefix for namespaces are not allowed in XPath. Please define a valid prefix.

The XPath filter checkbox is checked

The xPath query is set to /dns:drugs

Can you tell me what valid Prefix and Namespace I should be using?

As supplied in the example the Prefix and Namespace are empty.

 

If I uncheck the XPath filter checkbox the node runs but then on executing the Extract Drugs metanode I receive a warning:

WARN XPath Errors loading flow variables into node : Please set a name for the new column.

The nodes continue to run but I get lots of errors.

Is this most probably a function of the original misconfiguration?

Many thanks,

Mike

Hi Mike,

the following image shows the settings dialog of the XML reader for the DrugBank flow. The namespaces table should not contain any row not even an empty row.

Please notice, that the XML file from DrugBank must be extracted and the fileUrl variable must be empty in the Flow Variables tab.

Bye,

Tobias

Hi everybody,

I have tried to remake this example by downloading DrugBank XML file on http://www.drugbank.ca/downloads. Your example WF doesn't work on the last version of DrugBank.

Therefore, I attempted to change there nodes settings in order to extract data as discribed in https://www.knime.org/files/nodedetails/_xml_XPath.html and http://www.w3schools.com/XPath/xpath_syntax.asp. Unfortunatly, my experiment is unconclusive.

I'm a newbie with regard to this KNIME utility. I would like to learn how to configure settings of XML reader and XPath properly. I don't understand the utility of namespace and how to configure that.

Thank you in advance for your time,

Have a nice day :)

Baptiste

 

Configuration :

system : CentOS6.5

Knime : KNIME 2.11.2

Hi Baptiste,

I have adapted the exising workflow to work with version 4.2 and attached. We are also working on multi XPath extractor node that will reduce the size and improve the speed of the data extraction step drastically. Once the node is available I will update the example workflow on the server to work with 4.2.

Please notice that you have to unzip the XML prior processing it with the XML Reader node.

Bye,

Tobias
 

Hi Tobias,

Thanks a lot.

Baptiste

Hi Tobias,

 

Drugbank released a new version 5.0.0 a few days ago and I have the same problem as Baptiste had in the past: the xml reader node gives an empty table. Could you tell me what to change in Drugbank_Network_analysis to be able to read this new drugbank.xml file?

 

Thanks a lot in advance!

 

Peter

Same problem here. 

I think it is a problem with a mismatch between the xml file and the Schema. If you replace the second line of the Drugbank.xml file

<drugbank xmlns="http://www.drugbank.ca" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.drugbank.ca http://www.drugbank.ca/docs/drugbank.xsd" version="5.0" exported-on="2017-04-01">

For just

<drugbank>  

 

then it will work. Of course, you will have to remove the option to " Incorporate name of the root " 

Hi Tobias - the example workflow from the knime hub (05_DrugBank_Network_analysis) gives the error message regarding deprecated nodes

. Also gives an empty table. I therefore imported the workflow 01203001_drugBank_4_2 - but there are deprecated nodes (e.g. missing values) which stop it working. Is there a workflow available that is currently working? Many thanks, Shoumo

Hi @Shoumo -

To get the Palladian nodes, you must now use a separate update site as described by @qqilihq in this post. Note that you should change the update site URL to match the version of KNIME you’re using (probably 4.1):

This should get you the required nodes, with the exception of the NekoHtmlParser. I wasn’t familar with this node, but I poked around on Nodepit and found that using the HtmlParser node is now recommended as a replacement:

Maybe Philipp can tell you more.

(On a side note, note that in most cases using deprecated nodes are fine - they should still work, it’s just they’ve been replaced by more up-to-date functionality. I think the main issue in the case of this workflow is not that the nodes are deprecated, but that the extension containing them moved to a different update site.)

Hope this helps!

3 Likes

Thanks Scott - that is most useful to know. However when I try and install Palladian (or anything else from NodePit) I come up with a KNIME security warning that these are “unsigned”

and the authenticity cannot be validate. How do we deal with this? Regards Shoumo

If you want to install them, you’ll have to click on “Install anyway”.

We deliberately do not sign our plugins, due to the considerable additional overhead and complexity.

– Philipp