0000-0002-9890-5451The potential of using machine learning techniques in medicine is immense. As electronic health records have become widely available, there is hope that machine learning will improve diagnosis and care. However, integrating these new methodologies
Working together to improve the user experience: a conversation with Susan Stearns of the Boston Library Consortium
Q. Can you tell us about your background and your current position? A. I’m a librarian by training and worked early in my career in both academic and corporate libraries. However, the bulk of my career has been spent on the library software vendor side, working for companies such as Faxon Research Services, NorthernLight Technologies,…
18,000 years ago, the remote Indonesian island of Flores was home to a population of tiny humans. They stood only about 3.5 feet tall on their large feet, and their skulls housed unusually small brains approximately the size of a grapefruit. The identity of these ‘hobbits’ has been hotly debated for years: Were they modern humans suffering a disease, or a new species, Homo floresiensis?
Biological anthropologist Karen Baab first studied a model of LB1, the only skull recovered from the site, at the American Museum of Natural History in 2005. In a recently published PLOS ONE study, she and other researchers compare this specimen to a range of other modern human and extinct hominin skulls to get closer to settling the identity of Homo floresiensis, or ‘Flores man’.
The origins of ‘Flores man’ have been debated for quite a while now. What are the possible origins that are being discussed, and why the uncertainty?
The primary debate has centered on whether LB1 (and possibly the other individuals found on Flores) represents a new species that descended from an extinct species of the genus Homo or whether it is instead a pathological modern Homo sapiens, i.e the same species as us. If the Flores remains do in fact represent a distinct species, then the next question is whether they descended from Homo erectus, a species that may be our direct ancestor, or an even more primitive species. The latter scenario implies an otherwise undocumented migration out of Africa.
What makes it so hard to settle the argument one way or the other?
One of the difficulties in settling this particular argument is that most studies have focused on one or the other of these ideas and compared the Flores remains to either fossil hominins or to pathological modern humans, each using a different set of features. This makes it challenging to compare the alternative hypotheses side-by-side.
What kind of diseases might have caused modern humans to have features similar to these ‘hobbits’?
The three that have been discussed most prominently (and the three we looked at) are microcephaly, endemic hypothyroidism (“cretinism”) and Laron Syndrome. Microcephaly is not a disease per se, but rather a symptom of many different disorders. It refers to having an abnormally small brain and therefore skull. “Cretins” suffer from a lack of thyroid hormone before and after birth, which leads to stunted growth and possibly a slight decrease in skull size. Laron Syndrome individuals produce growth hormone, but their bodies do not properly recognize it, again leading to stunted growth and other developmental issues.
Only a few specimens of this hominin have been found, and there’s only one known skull, from the specimen named LB1. Are there reasons why these specimens have not been discovered elsewhere?
If Homo floresiensis descended from Homo erectus, then their closest relative lived just “next door” on the nearby island of Java. In this case, the unique features of the Homo floresiensis species probably evolved in the isolated island environment of Flores. If, however, the ancestor was a more primitive species, and Homo floresiensis didn’t branch off from H.erectus, it is possible that they might have migrated earlier than known, and we could still find older sites in mainland Asia containing this ancestral species.
You compared the morphology of the LB1 skull to many hominin ancestors and modern human populations from around the world. What were some of the most striking similarities and differences?
The LB1 skull is very distinct from the typical modern human’s, as it has a lower, more elongated silhouette when viewed from the side, , greater width at the rear of the braincase, and a flatter frontal bone (the bone underlying the forehead) with a more pronounced brow ridge. Interestingly, these are some of the same features that distinguish archaic species like Homo erectus from modern humans.
Specimens of Laron Syndrome and “cretin” skulls from modern Homo sapiens presented large, round, globular braincases, which are very different from the smaller, lower and less rounded braincase of LB1. The microcephalic human skulls present a closer comparison to LB1, but still show clear distinctions from LB1 in much the same way that they differ from species like Homo erectus or Homo habilis.
Overall, the LB1 braincase is most similar in its overall shape to small-brained Homo erectus from Eurasia that are 1.8 million years old.
How does this analysis add to, or change, what we knew about Flores man?
This analysis provides a unique opportunity to evaluate these evolutionary and pathological hypotheses side-by-side based on the same criterion – of cranial shape similarity. The results support a stronger affiliation of LB1 with fossil Homo than with any of the proposed pathologies. This study also offers an improvement over previous assessments of the microcephaly hypothesis by using a more extensive sample that better captures the variability in this disorder.
Do these results conclusively settle the discussion? What other possibilities still exist for the origins of H. floresiensis?
While very little in paleoanthropology is ever “settled,” I do think this study represents an important step forward in terms of putting the pathological hypotheses to rest. The question that remains to be answered definitively is which species of archaic Homo is the most likely ancestor of Homo floresiensis – Homo erectus or an earlier and more primitive species of Homo?
Citation: Baab KL, McNulty KP, Harvati K (2013) Homo floresiensis Contextualized: A Geometric Morphometric Comparative Analysis of Fossil and Pathological Human Samples. PLoS ONE 8(7): e69119. doi:10.1371/journal.pone.0069119
Small pieces of RNA in our cells can act like molecular switches that turn genes off by binding to them. These pieces, called small interfering RNAs (siRNAs) are also used by researchers to design experiments to understand what certain genes do.
Scientists can design siRNA molecules aimed at turning off specific genes they are trying to study. Though such siRNA ‘switches’ can be very useful, they are often non-specific, turning off hundreds of genes that they should not have an effect on. As a result, it is difficult for a biologist to conclude whether an experimentally observed effect is due to turning off the gene they meant to turn off or the hundreds that they didn’t (called “off-target effects”). Eugen Buehler of the National Center for Advancing Translational Sciences (NCATS), a new center at the NIH, describes an alternate approach to dealing with these off-target effects of siRNAs in his recent PLOS ONE paper. Read on to find out more about this research:
How did you become interested in improving siRNA experiments?
I’ve been working on siRNAs for about the last five years. When I would talk to my wife (a cell biologist) about my research, I would go on and on about all the problems created by these non-specific effects. Since she uses siRNAs in her research, she would ask, “Well, what should I do to avoid it?”
I didn’t have an answer. All the methods I had involved a statistical analysis of a large number of results from high-throughput screens, which look at several thousand genes at once. They couldn’t be applied to experiments that only involved one or a few genes, which is what many researchers do. It frustrated me not being able to help her, and so this question of what to do about off-target effects in small-scale experiments kept nagging me, until I found an answer.
And what was that answer?
As is often the case, the answer involved looking at the problem a different way. For years, people have been trying to solve the problem by getting rid of the non-specific effects. There are many ways to do this, but they still have a high incidence of these effects.
So, rather than trying to eliminate the off-target activity, we took the opposite approach. We changed three points (bases 9-11) where an siRNA makes contact with its target, so it couldn’t have the effect it was designed for. In this way, we created the C911 version, an anti-siRNA of sorts, which had all the off-target effects but none of the on-target effects.
So if an siRNA has an effect in a cell that is different from what the C911 version of the same siRNA has, we can conclude that the effect is because it silenced the intended target.
Which figure in the manuscript do you think best summarizes your results?
Definitely Figure 3B. Here, we compare siRNAs that appear to have specific effects but don’t (false positives), with ones that do have a specific action on a target gene. We took ten of each kind and created C911 versions for all twenty.
When we compared the two we found that for the false positives, the siRNA and the anti-siRNA had the same effects (left hand panel). But for the siRNAs which really did have an effect on their target, there was a big difference between the siRNA and its C911 version. As it happened, the C911 controls worked perfectly for all twenty siRNAs we had selected for the experiment.
Where do you hope to go from here?
For a tool as well established as siRNA, it will take a while to change the way we design our experiments. The first step is for there to be a reasonable alternative, and that is what this paper is meant to supply. The next is to make that alternative easy to choose. Part of that will involve getting companies that manufacture siRNAs to eventually make negative controls like this, so that negative controls like C911 can be easily and affordably obtained for any siRNA.
My hope is that someday when a researcher orders an siRNA, they won’t even have to ask; they’ll get a tube with their siRNA and a tube with the appropriate negative control by default.
Citation: Buehler E, Chen Y-C, Martin S (2012) C911: A Bench-Level Control for Sequence Specific siRNA Off-Target Effects. PLoS ONE 7(12): e51942. doi:10.1371/journal.pone.0051942
Image: Target by Ivan McClellan on flickr