Interactive Bioactivity Prediction with Multitask Neural Networks

A CHEMBL-OG publish, Multi-task neural community on ChEMBL with PyTorch 1.0 and RDKit, by Eloy, from method again in 2019 confirmed easy methods to use information from ChEMBL to coach a multitask neural community for bioactivity prediction – particularly to foretell targets the place a given molecule is likely to be bioactive. Eloy has hyperlinks to extra data in his weblog publish, however multitask neural networks are fairly attention-grabbing as a result of the way in which info is transferred between the totally different duties throughout coaching may end up in predictions for the person duties which can be extra correct than what you’d get for those who simply constructed a mannequin for that process alone.

It’s a giant distinction to most people: Our efficiency tends to go down the second we begin multitasking. In any case, I discover this an attention-grabbing downside and Eloy offered all of the code essential to seize the information from ChEMBL and reproduce his work, so I made a decision to select this up and construct a KNIME workflow to make use of the multitask mannequin. For as soon as I didn’t should spend a bunch of time with information prep (thanks, Eloy!), so I may immediately use Eloy’s Jupyter notebooks to coach and validate a mannequin. After letting my workstation churn away for some time, I had a skilled mannequin able to go; now I simply wanted to construct a prediction workflow.

Loading the Community and Producing Predictions

Eloy’s notebooks construct the multitask neural community utilizing PyTorch, which my firm’s platform doesn’t immediately help, however fortuitously each platforms help the ONNX (Open Neural Community Alternate) format for interchanging skilled networks between neural community toolkits. So I used to be in a position to export my skilled PyTorch mannequin for bioactivity prediction into ONNX, learn that into my firm’s platform with the ONNX Community Reader node, convert it to a TensorFlow community with the ONNX to TensorFlow Community Converter node, and generate predictions utilizing the TensorFlow Community Executor node. 

Now that I’ve the skilled community loaded into the platform, I must create the right enter for it. Because the mannequin was skilled utilizing the RDKit, that is fairly simple utilizing the RDKit KNIME Integration.

I do know that the mannequin was skilled utilizing the RDKit’s Morgan fingerprint with a radius of two and a size of 1024 bits and I can generate the identical fingerprints with the RDKit Fingerprint node. Since I can’t cross fingerprints on to the neural community, I additionally add an Increase Bit Vector node to transform the person bits within the fingerprints into columns within the enter desk. The compounds that we’ll generate fingerprints for are learn in from a textual content file containing SMILES and a column with compound IDs that we’ll use as names. The pattern dataset used on this weblog publish (and for the instance workflow) is made up of a set of molecules exported from ChEMBL and a few invented compounds I created by manually enhancing ChEMBL molecules.

The output of the TensorFlow Community Executor node is a desk with one row for every molecule we generated a prediction for and one column for every of the 560 targets the mannequin was skilled on. The cells include the scores for the compounds towards the corresponding targets, Determine 2.

At this level now we have a fairly minimal prediction workflow: We are able to use the multitask neural community to generate scores for brand new compounds. In the remainder of this publish, I’ll present a few methods to current the outcomes in order that it’s a bit simpler for folks to interactively work with them.

Displaying the Predictions in an Interactive Heatmap

The primary interactive view that we’ll use to show the predictions from the multitask neural community features a heatmap with the predictions themselves and a tile view exhibiting the molecules the predictions have been generated for. The heatmap has the compounds in rows and targets in columns with the coloring of the cell decided by the computed scores. The tile view is configured to solely present the chosen rows.

The “show predictions as heatmap” element that exposes this interactive view is ready up in order that solely chosen rows are handed to its output port. So, within the instance proven in Determine 3, there would solely be two rows within the output of the “show predictions as heatmap” element.

The workflow does a big quantity of information processing with a purpose to assemble the heatmap. I received’t go into the main points right here, however the principle work happens within the “reformat with bisorting” metanode, which reorders the compounds and targets primarily based on their median scores. This brings targets which have extra high-scoring compounds to the left of the heatmap and compounds with excessive scores towards extra targets to the highest of the heatmap. Qualitatively the heatmap ought to get redder as you pan up and to the left and extra blue as you pan down and to the suitable. There’s no greatest reply as to one of the best sorting standards for this goal, so be at liberty to mess around with the settings of the sorting nodes within the “reformat with bisorting” metanode for those who’d prefer to attempt one thing aside from the median.

Evaluating Predictions to Measured Values

A good way to achieve confidence in a mannequin’s predictions is to check them with measured information. Usually, we are able to’t do that, however typically there will likely be related measured information accessible for the compounds we’re producing predictions for. In these circumstances, it might be nice to show that measured information along with the predictions. The rest of the workflow is there to permit us to do exactly that, Determine 5.

This begins by producing InChI keys for the molecules within the prediction set, wanting these up utilizing the ChEMBL REST API, after which utilizing the API once more to seek out related exercise information that was measured for these compounds. Daria Goldman wrote a weblog publish, titled A RESTful Approach to discover and retrieve information a few years in the past, exhibiting how to do that. I’ve tweaked the parts she launched in that weblog publish for this use case and mixed all the things within the “retrieve ChEMBL information when current” metanode.

The output desk of the metanode has one row for every compound, a ChEMBL ID for every compound that was present in ChEMBL, and one column for every goal the place there was experimental worth in ChEMBL for one of many compounds within the prediction set. This information may be visualized, along with the predictions utilizing the “Show predictions and measurements” element, Determine 6.

This interactive view is based on the scatter plot on the high. Every level within the plot corresponds to 1 compound with information measured towards one goal. The CHEMBLIDs of the targets are on the X axis and the measured pCHEMBL values (as offered by the ChEMBL internet companies) are on the Y axis. The dimensions of the factors within the plot is set by the calculated rating of that compound for that concentrate on. The scatter plot is interactive: deciding on factors exhibits the related compounds within the desk on the backside left of the view and the corresponding scores and measured information within the desk on the backside proper.

If the mannequin is performing very well, I’d count on the scatter plot in Determine 6 to have massive scores (large factors) for compounds which have excessive exercise (massive pCHEMBL values), i.e., larger factors in direction of the highest of the plot and smaller factors in direction of the underside. That’s roughly what we observe. There are clearly some outliers, however it’s in all probability nonetheless OK to pay not less than some consideration to the mannequin’s predictions for the opposite compounds/targets. (Notice: This isn’t a totally legitimate analysis since a lot of the information factors I’m utilizing on this instance have been really within the coaching set for the mannequin. The instance is proven right here with a purpose to show the view and its interactivity.)

Wrapping Up

On this weblog publish, I’ve demonstrated easy methods to import a multitask neural community for bioactivity prediction constructed with PyTorch right into a workflow after which use that to generate predictions for brand new compounds. I additionally confirmed a few interactive views for working with and gaining confidence within the mannequin’s predictions. The workflow, skilled mannequin, and pattern information can be found on the hub so that you can obtain, be taught from, and use in your personal work.

Obtain the “generate predictions with ONNX” workflow from the hub right here.

As first revealed on the KNIME Weblog.