PhD position
Postdoc position
So, there we are. This is where spreadsheets come in. If done well, they combine aspects of knowledge graphs with usability and can even cover a good bit of the learnability. This is what is described in this new paper about templates in the EU NanoSafety Cluster: A template wizard for the cocreation of machine-readable data-reporting to harmonize the evaluation of (nano)materials (doi:10.1038/s41596-024-00993-1).
The learnability comes in with the spreadsheet templates (“this is how we did it”) and a “wizard” around it guides the user with the selection of a template but also can provide feedback on the template. The technical term for that is “validator”, but it can be tought of as a spelling checker. Computers are good at finding contradictions (the lack of a pattern), though less good at ranking the alternatives (which is the cause of hallucinations in AI approaches).
And to return to the RDF, software like AMBIT can read these templates, use the semantics linked to the template, and make the FAIR static spreadsheets (good for archiving on Zenodo!) available as FAIR interactive data (good for exploration and machine learning), and as RDF (good for data integration).
Congrats to Nina and the various EU NanoSafety Cluster projects!
]]>PhD position
Postdoc position
// requires an unspecified Bioclipse
// development version
bioclipse.requireVersion("2.6")
service =
"http://apps.ideaconsult.net:8080/ambit2/";
serviceSPARQL =
"http://apps.ideaconsult.net:8080/ontology/";
stringMat = opentox.listDescriptors(serviceSPARQL);
stringMat.getColumn("algo");
stringMat.getColumn("desc");
// pick any descriptor
descriptor = stringMat.get(1,1);
opentox.calculateDescriptor(
service, descriptor,
cdk.fromSMILES("CCC")
)
The first descriptor happens to be a model for predicting the pKa (see Algorithm or Model: OpenTox API quiz).
]]>As, a result, you can (with the code at my laptop) now do (see this BSL script at MyExperiment):
// requires an unspecified Bioclipse development version
ds = opentox.createDataset("http://apps.ideaconsult.net:8080/ambit2/");
opentox.addMolecule(ds, cdk.fromSMILES("CCCCC[N+](C)(C)C"))
opentox.addMolecule(ds, cdk.fromSMILES("ClC(I)Br"))
opentox.deletaDataset(ds);
Make sure to check out the other stuff I have been doing with respect to OpenTox.
]]>The metabolic fate of molecules can be predicted with the MetaPrint2D feature (see also the main MetaPrint2D page). The feature currently uses one specific method, but the immediate feedback you get while drawing molecules, can basically use any model. The currently used model is developed in a collaboration of Ola with Sam Adams (whom we all know for his JNI-InChI library) at Cambridge University and Lars Carlsson at Astra-Zeneca.
This screenshots shows the visual feedback in the (new) JChemPaint editor in Bioclipse, and on the right we see one interesting Bioclipse feature in action: the cheat sheets. These cheat sheets are inline help documentation which guides the reader through the functionality. But, unlike mere help, cheat sheets can be very interactive and perform some tasks itself, making it easier for the reader to see what was supposed to happen.
L. Carlsson, O. Spjuth, S.E. Adams, R.C. Glen, S. Boyer, Use of Historic Metabolic Biotransformation Data as a Means of Anticipating Metabolic Sites Using MetaPrint2D and Bioclipse, accepted in BMC Bioinformatics.
Structural alerts are one method to signal the scientist that the molecule under study needs some more attention (see for example doi:10.1016/j.mrrev.2008.05.003). It helps him decide to continue to look at that particular structure, or to move on. Ola also developed a decision support plugin.
One of the cool features is that, in good Bioclipse habits, deliver a pluggable architecture. This practically means, that anyone can add their own decision rules; those can be added as local software, or as services on a central, institute specific server. The results in this screenshot show that a Bursi AMES data-based model estimates that this molecule is mutagenetic.
O. Spjuth, L. Carlsson, M. Eklund, E.A. Helgee, S. Boyer, Integrated decision support for assessing chemical liabilities, In preparation
]]>
More on the meeting later.
]]>
Because the virtual talk required me to reboot into Windows, I could not give a live demo of the development version of Bioclipse, showing new stuff in action. The audience had to with just screenshots :( I’ll try to make a screencast soon.
The Bioclipse OpenTox plugin currently has this functionality:
]]>The meeting is also bound to be fun. I have not done much in the area of toxicology other than the more general QSAR/QSPR model building with chemometrics. But I have been recently taking to Nina and other of the OpenTox community, and started to play a bit with the data and computation API they are developing.
I started a Bioclipse plugin recently (see screenshot), and placed the source code in this bioclipse-opentox Git repository on Gitorious (my GitHub account is already over the formal limit). The functionality is still quite limited, and the manager currently only provides methods to download data sets (myexperiment:1192):
// query a service using the OpenTox API 1.1
// See: http://www.opentox.org/dev/apis/api-1.1
var service = "http://apps.ideaconsult.net:8180/ambit2/";
var datasets = opentox.listDataSets(service);
for (set=0; set<datasets.size(); set++) {
var dataset = datasets.get(set);
js.say("Downloading set: " + dataset);
ui.open(
opentox.downloadDataSetAsMDLSDfile(
service, dataset, "/OpenTox/ambit" + dataset + ".sdf"
)
)
}
Behind this plugin is again the RDF plugin, as OpenTox uses RDF too, a few simple SPARQL queries was all that needed to be defined. And again , the Bioclipse pluigin code base is pretty small.
]]>