The engine is innovative in that it not only will index the materials used in every experiment, it will also track the frequency and rate at which they are used
Palo Alto-based Bioz, a search engine that indexes and tracks trends in the sciences by algorithmically analyzing research papers, has raised $3 million, the company announced Wednesday morning. The round was led by life sciences-focused 5AM Ventures with participation from the Stanford StartX Fund, Astia Angels, investor Esther Dyson and others.
The engine is unique relative to other science-related search engines like Sparrho in that it not only will index the materials used in every experiment, but also track the frequency and rate at which they are being used throughout life science research worldwide.
The company was co-founded in 2015 by Chief Scientific Officer Karin Lachmi, Ph.D. and and serial entrepreneur CEO Daniel Levitt. After less than a year in operation, word of mouth across the academic landscape has garnered it 30,000 registered users from more than 1,000 universities and companies in the biopharmaceutical industry.
“I still didn’t have the basic internet tool,” Lachmi reminisced to Geektime, frustrated by having to do intense research and still not having a more calibrated tool given the ubiquity of advanced search platforms. Lachmi is based at Stanford, where she has already completed two postdocs after moving there from Israel. “I wanted an engine that would help me choose what products I need in my experiment, what would work, how to work with them, and how long should I leave them.”
Levitt is no stranger to the sciences himself. His father, Dr. Michael Levitt, is a Nobel prize winner in chemistry.
“There are 20 different categories, each one with 300 million products or 10,000 vendors. I think we have 200 million different product pages and 500 million insights from the articles,” he accounts. You can see “which products used together, which products in different types of experiments. Once we find it in one article we see where it appears in other articles as well.”
The search also debuts the Bioz Star rating system with the announcement, which ranks products based on frequency of mention, impact factor (citation index) based on authority, and the recency of publications that feature those research materials. They also rate according to 1) efficacy factor, or how the result fairs against other results; 2) tendency factor, comparing the result to the average of other related experiments; and 3) protocol relevance, or how often the result appeared with a protocol condition.
This all helps researchers find the items that suit their specific assays, or their particular experimental procedures.
A gigantic industry with no one to organize it
According to Bioz’s founders, upwards of $80 billion is spent annually on instruments and biological material for research. Perhaps half of that purchased material, antibodies for instance, is not suitable for a given line of research or won’t result in a successful experiment. This search index looks to optimize purchasing by demonstrating what lines of products are trending in a given industry and line of research.
Lachmi and Levitt described one user who had been researching what sort of antibodies might be suitable to fight a specific kind of infection. The conclusion he reached over the course of six months after testing out several was the same that Bioz came up with in a mere six seconds based on its assessment of antibodies being used in the field for similar research.
“Imagine, six months just because they used the wrong agent,” Lachmi remarked.
And by the way, Bioz is free to use.
This tool makes Google Scholar look like an elementary school library in comparison. Scholar is notoriously difficult to navigate and data was hardly tailored for any specific sector of academia, let alone the sciences. The books and papers are also not free to look at, which is an anathema considering many scientific papers are supported by public funding.
“Typically, money for research comes from the government. At the end of the day, the National Institute of Health (or NIH, which is behind many public grants) is taxpayer money. There’s free access to a huge data set.” And that is just in the United States. The EU, by 2020, plans to make all scientific papers publicly available.
Other results could be found through the data, including primary public resources for funding, leading or upcoming researchers in the field, and obviously speeding up the process of drug discovery.
According to Lachmi and Levitt, a new paper is published somewhere in the world about every 10 seconds. They are usually 30 pages in length, clearly creating a demand for this sort of algorithmic approach to sorting the data.
When asked if there was any concern that the results of searches might be skewed by corporate-sponsored research papers, Levitt said that over 90% of the papers were from academic sources and played down concerns. The sources of the papers are also clearly marked in the results.
The natural language processing power necessary for this is immense, as it requires correlating dozens of data points and being able to charge the evolution of how, by whom and when certain products are used.
A hub for product referrals, an Au mine
Bioz’s business model is also painfully simple: they get referral payments for click-throughs to vendor websites. When materials are mentioned in given papers, users have the option to immediately go to the online store and page where it’s on sale.
Is it a challenge to sign up thousands of vendors? Not really, says Levitt. About 60% of the market is dominated by five or six companies anyway, and from there onward it becomes an arms race. No smaller vendor wants to give a competitive advantage to a competitor who might sign up for the referral program with Bioz. No one has pushed back, say Lachmi and Levitt.
One drawback might be the time between experiment and publication, which may be several months. That being considered, their trend charts will still be the most up to date, as there is as yet no other easy way to track purchases industry-wide or mark other movements and course corrections among researchers.
“We’re talking about adding filters, like to say a certain person made this product within a certain budget. We know from the NSF (National Science Foundation) data what grant everyone got and to which you can tie to a specific researcher. But we know exactly what each person bought.”