ReadCube, the reference manager software from Nature Publishing Group, has recently released major update/re-development, which finally makes it a usable gadget and a legitimate competitor of Mendeley, my favorite one in this category.
Since my early days in science (2008 ish), I’ve always wanted a perfect file manager that organizes and displays my pdfs——-something like the “Papers” on Mac, but works on PC, has a cloud storage and is free.
After several months’ searching and comparing, I settled down on Mendeley and have used it ever since.
Mendeley has everything I need. Concise UI, automatic metadata fetching, great PDF display, notes, highlights and tags. The free version offers 1G (2G now) of cloud storage, which is a decent amount for PDF files, and sync between the desktop version and the web. It also has a web importer bookmarklet which allows instant paper import to the library when the metadata is able to be extracted from the current webpage. This works on PubMed and most of the journal websites. It has something beyond my expectation. It introduces the “watched folders”, in which every PDF stored are automatically listed and updated in the software library. Even better, you have the option to let Mendeley rename your files based on key metadata information such as “author”,”journal” and “year”. Last but not least, it’s also an Endnote. Easy reference import and bibliography generation in office word, open office and libre office.
The only two minor flaws in my opinion is: the font size of the library display is too small (~10.5 pt) and cannot be changed, which makes it hard to browse from a 15” laptop. This later becomes less of an issue when I more and more rely on searching with tags and authors and basically not use the main library display anymore. The other downside of Mendeley is the memory allocation. I’ve experienced, when the library contains over 1,000 PDF files, severe lag and poor performance of the software. Since then, I reduced the number of actual PDFs saved on my disk, instead saved most of them only in metadata format.
Back to ReadCube. ReadCube was first introduced, I believe, sometime in 2012. I gave it a try because of its stunning UI and the ability to fetch reference hyperlinks and supplemental information from the paper. The user interface is very Mac like: simplicity, nice button effects and awesome colors. For some of the papers that are “enhanceable”, ReadCube can automatically extract reference and supplemental information from the paper, which is great. But ReadCube at the time did not offer many other useful features like what Mendeley does: web importer, write-and-cite tool, cloud and file renaming. What was the worst and essentially prevented me using ReadCube was the display: it has three columns, the library info panel on the left, references/abstract/supplemental on the right, and the actual PDF in the middle. The two panels on either side cannot be closed/hidden and therefore the effective reading space is greatly reduced, and on my 15.5” laptop, too small a font to read. You can zoom in, but everything became a mess to navigate. Also, the addition of references and supplemental often times makes the software running slow and easy to crash.
But everything’s changed in this recent major re-make from last week. The two cumbersome panels can be removed during reading mode, which makes the actual size of reading space equivalent to that of Mendeley’s. Everything is smoother and quicker. I can finally enjoy reading a paper without having worry about the font size and potential crashes. It also introduces web importer and citation tool for Word. On the other hand, you have to pay for cloud space (though unlimited) and watched folder option for $5/mo. And it still doesn’t have tags and file-renaming. And for some reason (perhaps because of my customized font rendering), fonts in ReadCube do not look as sharp as in Mendeley.
So, Mendeley is still my go-to reference manager and paper organizer, although ReadCube has had some great features and improvement. Something I think could be useful is the online Mendeley community. You can invite friends and colleagues and share ideas and papers; you can see other people’s recommendation and comments on papers, sort of like F1000, but clearly less matured. I haven’t tested either of those because I have only one friend in my Mendeley account and we haven’t talked since February of 2011.

I was about to write something about this study led by Guri Giaever’s group from my beloved U of T MoGen department, when I saw the update from Derek Lowe’s blog on the same subject, where he and a few fellow chemists have spotted some silly chemistry mistakes.
First, they appear to recognize N-phenylbenzylamine and it’s biologically equivalent tautomer as two chemicals and apparently target two distinct cellular pathways.
Then there’s this issue with imidazole and protonated imidazole, which shouldn’t exist at all in a physiological buffer system.
And this one is particularly funny: they have this cyclohexa-2,4-dien-1-one

which should just be phenol indeed.
And there’s a few more examples of mis-nomenclature and small molecules invalid in a biological system like those.
What a shame on the authors! What a shame on the reviewers! and on Ron David, who is in the authorship list and whom I respect a lot.
PS. a note on Guri. Apparently she’s not at U of T anymore. She and her husband, another MoGen faculty, Corey Nislow have joined UBC since last year. Interesting move.

This paper is interesting. It’s not substantial.
I take it as a grain of salt. No, I don’t like it at all.
They were seeing reduced level of an metabolic enzyme Cystathionine γ-lyase (CSE), which makes cysteine and hydrogen sulfide in three models of huntington’ disease: cell lines, mouse models and post-mortem patient samples. And CSE KO mice exhibits certain behavior deficits resembling HD symptoms. Replenishing cysteine in diet alleviates these behavior deficits and increases survival.
First of all, it’s an interesting observation of a reduced level of CSE in HD models——perhaps among hundreds of other random proteins. There is no evidence suggesting CSE activity loss or global reduction in cysteine level, in any way, contributes to disease onset/progression.
Actually they can’t tell if cysteine is the culprit here. CSE makes both cysteine and hydrogen sulfide, and we know that H2S is an important second messenger in signaling in the brain and periphery. That cysteine rescue experiment does not rule out H2S function in the disease state.
The biggest problem I have with this study is all the behavior assay. They were trying to use a few of the standard motor tests to model HD, which is just not nearly satisfying at all. Who knows what causes these deficits when you deplete one of the nine essential amino acids ? It totally doesn’t have to be, in any way, Huntington-dependent, and these tests are not at all good means to monitor HD pathophysiology.
What should they do? Well, since it’s just too vague a connection between CSE expression and some random motor behavior tests, anything in between is essential and the more, the better. At least they should’ve looked at some cellular phenotypes including cell death, inclusion body aggregates and neurodegeneration. Some metabolic profiling and behavior are also welcomed.