The top three news stories of the week, as chosen by our resident students. This week’s top stories include plans for open access on scientific publications, using AI for image analysis and new findings in mitochondrial inheritance.
By Eduardo Serna Morales
China backs Plan S for open access on science publications
On September 4th, research funders from 11 European countries launched a plan to make all publications resulting from their funding complete and immediate open access by 2020 (See a note from PlosOne here and from Nature News here). The European Commission and the European Research Council immediately backed this plan, while two of the world’s larger medical research charities, the Bill and Melinda Gates Foundation and the Wellcome Trust, joined the “cOAlition S” on November 5th (See the press release here).
Since then, a fierce debate has been going on between the backers and detractors of Plan S, especially from the big research publishing companies that defend their business model as the only way to secure high-quality scientific research on the future. Nonetheless, Plan S has received a huge boost by an open letter from 1778 researchers from all over the world (that can be found here) and a statement from Chinese librarians’ and funders on December 5th in support of their pledges for complete and immediate open access science.
Now that China, the Wellcome Trust and the Bill and Melinda Gates Foundation have made Plan S a global initiative, is possible that other scientific funders from all over the world join this initiative in the future. In spite of that, it’s difficult to predict if Nature, Science, Cell and other journals that control around 85% of scientific publications will change their access policies to comply with these new guidelines. Regardless of the outcome, the public debate about open access in science will be beneficial for the future of academia and science overall. No matter which side you are supporting, you can be part of the conversation on Twitter by using the hashtags #PlanS and #OpenAccess.
Using Deep Learning for image restoration in fluorescence microscopy
Nowadays it’s very common to find interesting stories in social media and newspapers about how Artificial Intelligence (AI) could completely change the future of humankind. Although we all know that companies like Google, Facebook or Amazon are using Big Data, Image Convolution and other AI approaches for their platforms, there are very few cases in which non-specialists users can take direct advantage of this tools for their own purpose.
This is also true in the biological sciences, where the recent boost of scientific research using AI approaches is the result of interdisciplinary collaborations with AI experts. But as the field advances, these tools will be within reach of more people in the future. That’s one of the reasons that makes the recent paper from a group of scientists at the Center for Systems Biology in Dresden published in Nature Methods so interesting.
They trained content-aware image restoration (CARE) networks to restore fluorescence microscopy images in tasks like image denoising, surface projection, recovery of isotropic resolution and the restoration of sub-diffraction structures. And on top of that, the best part is that all their image restoration methods are freely available as open source software in Python, FIJI and KNIME through their GitHub!
If you work using fluorescence microscopy, this could be a great tool to save a lot of time and work, enhance the quality of your images and solve technical problems related to laser power and resolution. For using it you just need basic knowledge of python and to follow the instructions at the CARE GitHub. In no time you could be training a convolutional neural network with your own data, leaving AI to do its magic with your images, and generate a personalized FIJI plugin for improving similar sets of data in the future with just one click. Welcome to the future.
The curious case of paternal mitochondrial DNA
We can find an exception to almost any rule, and this time was the turn for the textbook definition of mitochondrial DNA (mtDNA) transmission in humans. Mitochondria (and thus mtDNA) are transmitted to sequential generations exclusively through the mother after an active elimination of paternal mitochondria. This mechanism is central for some genealogical DNA testing and for inherited mitochondrial diseases.
It was previously known that most of us have around 0.1% of paternal mtDNA, which is considered as negligible, but a recent publication on the journal PNAS showed biparental transmission of mtDNA in 17 members in three unrelated multigeneration families. Where the levels of paternal mtDNA ranged between 24 to 76%. These results were confirmed by three independent laboratories and new blood samples to discard a false positive.
Although maternal inheritance remains as the dominant form of mtDNA transmission, this new discovery could bring an opportunity to better understand the molecular mechanisms of mitochondrial inheritance and could be key for new treatments to avoid inherited mitochondrial diseases.