New resource to challenge widespread poor practice in biomedical researchers’ use of chemical probes

Breast cancer cells stained for DNA (red), NFkB (green), and a reactive oxygen species probe (blue). Credit: Julia Sero for the ICR, 2011.

Details of a major new online resource designed to improve biomedical research by helping researchers choose the best chemical probes for their experiments have been published in the journal Cell Chemical Biology. The 'Probe Miner' resource allows drug targets to be validated much more effectively.

Scientists created the new resource after analysing almost 2 million chemical compounds that could be used to test the effects of inactivating a specific protein in cells – and finding that many were inadequate for the job.

Researchers across the world are using chemical probes that lack specificity for a single drug target and are often so broadly promiscuous that they are incapable of yielding any useful data.

There are worrying implications for the reliability of many published studies.

A new alternative to unsuitable search engines

Biological researchers are commonly unaware of the limitations of claimed chemical probes, employing search engines or commercial catalogues that fail to discriminate between high-quality probes and those that are badly flawed or out of date.

Probe Miner, a publicly available, free-to-use web-based resource, was created by scientists at The Institute of Cancer Research, London, and is the first publicly available, large-scale objective resource on chemical probes.

It scrutinises data on more than 1.8 million chemical compounds for their suitability as probes against 2,220 human proteins and potential drug targets – picking out the good, the bad and the ugly for all to see.

Probe Miner is used to objectively rank chemical compounds – employing criteria for quality that include potency, selectivity and cell permeability. It gives lower rankings to the less suitable or bad probes – such as ones that hit several targets at once – and even lower scores to the truly ugly ones, which have indiscriminate, promiscuous activity against very many targets.

Using the Probe Miner resource, the researchers concluded that only 250 human targets, or 1.2% of human proteins, have sufficiently effective chemical tools to be reliable for research applications.

Visit the chemical probes web resource – Probe Miner – released to the research community and led by Dr Bissan Al-Lazikani, Professor Paul Workman, Dr Albert Antolin and colleagues.

Read more

The importance of selecting the right chemical probe

Chemical probes, or tools, are small molecules that scientists use to understand how individual proteins function, both in healthy cells and in diseases such as cancer. 

These chemical probes play a crucial role in discovering new drugs – by demonstrating, or ‘validating’, that the function of a protein target is important in disease and also by acting as prototype drugs to show that it is technically possible to inactivate the target of interest with a small molecule.

But not all claimed chemical probes are of the same quality, and biomedical researchers need better information to avoid selecting flawed or out-of-date probes for their research.

The new Probe Miner database allows researchers to search and rank the best chemical probes to address the scientific question they are trying to answer – increasing the robustness of experimental findings, while reducing waste of resources.

The ICR researchers uncovered a vicious cycle where scientists might find a weak probe via a search engine, use it in their research and publish their results, which then increases the likelihood of that probe being found by others through search engines and the cycle continues.

This can cause potentially better probes to be ignored as they drop lower down the results of web searches than weaker, but more widely used and published about, probes.

How Probe Miner can help

Probe Miner, however, objectively scrutinises over 200 million experimental measurements to overcome this bias and will become more effective as more data are added on the potentially better but under-studied probes.

It has been estimated that spending £150 on a flawed chemical probe can cost the scientific community billions of pounds in wasted time and effort – and can delay the discovery and development for patients of innovative cancer drugs.

The Probe Miner project was funded by the European Union, the Wellcome Trust, Cancer Research UK and the ICR itself. Dr Albert Antolin, a Sir Henry Wellcome Fellow, carried out the much of the analysis research behind Probe Miner.

Probe Miner has a range of user-friendly displays that provide information about the number of chemical probes available for each target, specific features of each probe and any limitations.

The resource lists the chemical probes that meet a certain quality threshold, and names the top 20 rated compounds for the protein target of interest. For ease, Probe Miner is able to provide the objective analysis and ranking using pre-set weightings for various quality criteria – but the weighting of the criteria used can be customised according to individual researchers’ needs.

Today's publication builds on the earlier pre-print that first described this resource which has been greatly enhanced in response to user feedback. The paper describes the research that went into developing Probe Miner and also explains how to use the Probe Miner website, with additional information on the website itself and illustrative use-case examples.

Probe Miner is designed to complement the expert-curated database Chemical Probes Portal by providing large-scale, objective, data-driven information that will be continuously updated to keep track of fast-moving advances in the field.

The researchers stress that the synergistic nature of the two resources makes their use in combination especially powerful for selecting the best chemical probes. This is aided by easy-to-use links between the two resources.

Data on nearly 2 million chemicals

Dr Bissan Al-Lazikani, co-creator of Probe Miner and Head of Data Science at the ICR, said:

“Good chemical probes play a vital role in biomedical research – helping scientists to specifically switch off the functions of proteins of interest within cells, to precisely understand the jobs carried out by different proteins. But too many of the probes used in research are not of good quality, do not hit protein targets specifically, and end up producing misleading scientific results.

“Biomedical researchers often lack the information they need to select suitable probes and are stuck with no accessible way to evaluate them. Yet on the other hand, we have tens of millions of published experimental measurements that can help – if only researchers had a way to navigate through them.

“To create Probe Miner, we performed statistical analyses of a vast amount of data on nearly 2 million chemicals so that we could objectively assess every compound. The Probe Miner resource arms researchers with all the available information for each probe, highlighting supportive evidence and potential areas of error or bias.

“Having this information at their fingertips will empower researchers to make the best, most informed decisions.”

Detailed, up-to-date information

Co-creator Professor Paul Workman, Chief Executive of the ICR, said:

“The field of biomedical research has a very serious problem with chemical probes – contributing to the current crisis in reproducibility and robustness. A high proportion of chemical compounds claimed as useful probes in fact hit multiple targets at once, and some are frankly disastrous – producing results which can be highly misleading or plain wrong.

“There are many so-called probes still in use that are now well known to experts to be deeply flawed, and yet have been applied in literally millions of publications worldwide and continue to be employed today.

“Our new Probe Miner resource provides biomedical scientists with detailed, up-to-date information about all chemical tools available, so they can make informed decisions about which ones are the best fit for their research. We hope this will make the results of research more correct and robust while reducing waste – of time and money.

“I would urge all biomedical researchers to ensure that the probes they are using are up to the job – because otherwise their science will suffer, vital drug discovery research will be suspect, and progress to patient benefit will be delayed.

“We were disappointed to find such a low number of quality chemical tools available to study human proteins and to see so many examples of poor practice. It is vital that we encourage the research community to continue to develop new chemical probes, if we are ever to move away from well-studied drug targets and truly innovate in drug discovery.

“But the probes we use must be of sufficient quality and fit for purpose or science and patients will pay the price. We think Probe Miner will be a big help.”

Use Probe Miner for free.