January 1, 1970 (Vol. , No. )
Zachary N. N. Russ Bioengineering graduate student UC Berkeley
Here’s a question: Which product requires minuscule amounts of physical capital, an even tinier marginal cost per unit sold with capacity to fill infinite demand, and can make you very wealthy almost overnight? If you said music or web content you’re right about the first two points, but it is software that really fits the bill.
Of the five wealthiest people in the world, two made their money on software, and the iPhone app market mixes good ideas with a bit of programming into hundred-thousand dollar jackpots. Software is the work-from-home, get-rich-quick scheme that actually works. Why would anyone want to give that away?
Opening the Pipeline
Open-source computing looks at software as a public good. Programs are measured in pride over profit. Programs are provided with their source code free-of-charge, allowing users to freely modify the work, provided the terms of the license (usually attribution and licensing the derived code under the same open-source license) are met.
For mainstream applications such as operating systems or productivity suites, open-source alternatives followed their commercial counterparts after the latter’s licensing burdens became too restrictive. However, a large fraction of academic projects have been open-source from the start, owing to their educational value, collaborative nature, and public funding. NCBI BLAST is one of the most successful of these projects and receives hundreds of thousands of DNA-alignment requests daily.
The Cost of Freedom
With open-source software allowing more uses of the code—freedom to modify, distribute, and use—why isn’t open-source software dominant? Or is it?
As it stands today there are open-source options for virtually every step in the drug-development pipeline. Odds are the company website is being delivered by open-source Apache (60% of market share). Even so, the workstations are probably running Microsoft Windows (80% do) with some combination of proprietary and open-source applications.
Users are familiar with Windows, and choosing Windows offers a greater selection of programs in most cases. Drug development is already an expensive endeavor, so paying a relatively small premium for additional functionality and familiarity can be well worth it.
In the lab, open-source software co-exists peacefully with commercial software. You might see a presentation in Powerpoint filled with pictures analyzed and modified in open-source ImageJ.
Academia seems to favor open-source more than industry. One good example is how open-source statistics package R is rising in academic citations and discussions but still remains absent from job postings.
Also, computationally focused work favors open-source as well: Freedom to modify is critical when the algorithm is as experimental as the data itself, as in bioinformatics. Some main complaints with open-source software in biology are difficult installations, poor integration between packages, and lack of support.
How Do You Sell “Free?”
The free-to-download, free-to-modify code makes for a challenging business model. Still, there is a market in offering a fix. Open-source vendors create a distinction between free and paid “Enterprise” versions using some combination of additional services such as technical support, quality testing, proprietary add-on utilities, certification, and training.
By including these perks, they ameliorate the shortcomings of classical open-source, for a price, usually a fraction of competing costs. This is the case with OpenClinica, a clinical electronic data capture solution. Everyone can download the same base software, but only paid users receive validation, hosting, support, etc.
Though this blurs the lines of open- and closed-source, the confusion doesn’t stop there. Other licenses such as for the molecular dynamics tool Amber are not free but do include the source; the prices are split between $400 per non-profit/academic site and $20,000 for commercial sites. Finally, premium tools such as Accelrys’ Discovery Studio integrate open-source molecular dynamics tool NAMD with closed-source tools together in one giant package.
The question of which software option to use comes down to two main questions: What are your employees familiar with, and what is their time worth? Cash-strapped academic labs have shown that open-source software options are generally up to the task, but tweaking, fiddling, fixing, and searching is often necessary to get them running properly.
Once an open-source tool reaches a threshold of usability, though, schools start to integrate it into curricula; a phenomenon that will only become more common as library and software budgets are slashed. OpenClinica generated interest in academic circles after it was featured in PLoS Medicine, and its free version was demonstrated in an independent investigator-run 400-patient study.
It’s already made its way into a course at the Bioinformatics Institute of India and been picked up by several CROs including Athena Healthcare in China, Qualitix in Taiwan, and Emphron in Australia. This is a small fraction of the market, but it’s a start, and as areas like China and India grow in wealth, so too will the value of those CROs. Open-source software could take over newer, smaller, and poorer CROs that have fewer patients and could expect a greater fraction of licensing overhead.
Even so, changing software is a relatively small modification. If you could afford Oracle Clinical or one of the other commercial packages, then clinical trial workflow will not be particularly altered. For those who couldn’t, however, data collection and dissemination will become much easier.
Open-source represents an answer to software piracy, price-gouging, vendor lock-ins, and abandoned code. It comes with just one question: Who wants to install it?