Monday, 15 February 2010

NCBI pubmed 'enhanced' web site with java script is pretty but is [a bit] wrong!

At least is wrong not to have the previous interface for no-javascript browsers.

[update] the interface seems better now, but still have some issues like:
  • - In the display settings, after selecting the options, the 'return key' goes to the search field not the apply button!!
  • - The same for the 'send to'. Also here the interface is a slow two steps. You need to click in the option and some extra info appears at the bottom.

  • - The thing that really needs to be improved is the 'collections' interface:
    • After selecting some entries, and 'send to collections' you go to the following interface:

    • You need to go to 'append to an existing collection' and select one. All this is a bit time consuming. I would prefer to have the collection integrated in the pubmed pages

The usability (quick access and no mouse access) is low with the new javascript interface. It has nice effects but the animations slow things and you cannot use tab/arrow to select things.

Also the myNCBI is a nice idea, but it is very slow to add entries to your libraries it is faster to use citeulike with the doi/pubmed selector bookmarklet. The send-to/collections is unusable if you want to add more than two references. You can spent the whole day waiting for pages and selecting things. It should have a list of my collections and send directly there (as when you send to the clipboard).

It could be an awesome enhancement for newbies surfers, but a bit of a hassle for power users.

The new interface is 'modern' but not yet very efficient.

I agree also with this post:

Friday, 12 February 2010

best practices for writing web forms

Today Caroline Harrett was giving a talk about web from at EBI campus.
Caroline Jarrett: Forms and their Users

Caroline Jarrett: Forms and their Users « EBI Interfaces

Confluence does not have sectional editing, Are you kiding? nop!!!

It's sad but today I realized that confluence does not have sectional editing and they have been discussing about that since 2005!!!!!

[#CONF-5913] Sectional Editing - Atlassian JIRA
Sectional Editing

They propose some workaround, but they are not good. Basically they use {include} to include pages but then, you miss the subsections from this pages in the main TOC.

Also this is distracting. Wiki (as aHawiian word) mean quick! and this is the thing that I want. Quick text editing low I like to edit the wiki in markup-txt and sometime I write my wiki pages in emacs and copy/paste to the wiki later. This 'include' workaround breaks one of the good things of the wiki: its flexibility to combine big pages and multilinked micropages. Sometimes you want one thing, sometimes you want the other. Big pages without the sectional editing is a a bad thing because:
  • a) you block the whole page for editing preventing other users to contribute
  • b) difficult to find your text in the only text area (and having confluence headings like h1. h2. instead = =, == == etc does not help)
  • c) is very slow to load a page
  • d) moving from wisiwig to wiki-markup puts you back at the beginning of the text and you need to find your place again

Monday, 8 February 2010

Nice article about how to make bioinformatic questions to perlmonks

Perl Monks meditation: RFC: Bioinformatics Tutorial by BioLion

This is a nice page to guide you how to make bioinformatics questions to programmers.

Sunday, 7 February 2010

blody scribefire: why are you inserting zamanta tracking services to my blogposts without warning!!!!!!

Short story:
WTF! I have discovered that scribefire add a tracking service at the end of all my posts without any explicit warning:
<div class="zemanta-pixie">
<img src="[long%20tracking%20number%20here]" class="zemanta-pixie-img" /></div>

This is an invisible image for tracking purposes. This is ON by default, but at least you can disable from options/publishing/Automatically insert invisible tracking pixel for statistics gathering.

The guys at Scribefire are not playing nicely!! They should advertise this thing explicitly and also in the option they should put a link explaining what this 'statistics gathering' is for.

-20 for Scribefire!!!!

Long story:
Scribefire is a useful blogging tool. But I don't trust third part plugins. So I put all of them in quarantine after installing them and after each update (you never know if other not-nice-company has bought them after your last check for niceness).

Scribefire was not different. First of all I don't trust them (not anyone else) to give them my google mail password in order to be able to upload the post. So I created a new google account only for blogging and give access to my main blog.

Also I don't trust the wysiwyg editors not adding rubbish to the html so I always edit in html.

The problem came when I needed to add an image: I selected to upload it with my google api and I was expecting the image to go to my 'picasa', but it is uploaded to a place that I don't know with no explanation:

<img src="" />

As I am paranoic with these things, I decided to upload the post as draft and add the photos myself in blogger. But When I edited in html the blog I saw this infamous Zamanta tracking image at the end!!!!

I saw that other blogs are reporting this since one year ago!!!

Please ScribeFire you have the rigth to make your software as you please (trakings, addware, advertising etc) but respect my rights of having an INFORMED DECISION on my choices. There are a lot of bad guys around doing nasty things, so good guys should be as open and clear as possible and you are not doing very well at that respect. Shame on you!

Also this is very annoying. My blogs are my blogs and you don't have rights to put things there without my permission!!!!

Why dynamite plots are BAD

Don't use dynamite plots (barcharts displaying mean and std desv). Boxplots and violin plots exist for a reason!!!

Please read the following article about the perils of interpreting dynamite plots

And here a more balance article

A and B seems the same, C and D are the same and A/B are different of C/D

A has a large distribution
B has only TWO POINTS!!
C has an outlier
D Two distributions??

== Example in R ==

[from the Paul Murrel R graphics book]

* Barplot:

* Boxplot:
** first the boxplot
** then draw the point with jitter (random shift from the x to avoid overlapping)

Saturday, 6 February 2010

Brian Kernighan talking about smart phones

What makes smart phones smart?
By Brian Kernighan GS '69

Arthur Clarke’s famous Third Law says, “Any sufficiently advanced technology is indistinguishable from magic.” Shazam is not magic: The various technological pieces are straightforward, though the specific algorithm that can so efficiently identify one tune out of zillions surely qualifies as advanced technology. Some deprecate the whole thing with comments like “It’s not really very good at classical music.” True enough, but viewed more generously, it’s nothing short of astonishing.

I kid my students about endowing a chair when they become rich and famous, but the real reward is watching enthusiastic and creative people convert dumb devices into smart ones. That really is indistinguishable from magic.

reading about indentation styles in wikipedia

Human population relationships from genome scans

Worldwide Human Relationships Inferred from Genome-Wide Patterns of Variation -- Li et al. 319 (5866): 1100 -- Science
Science 22 February 2008:
Vol. 319. no. 5866, pp. 1100 - 1104

Worldwide Human Relationships Inferred from Genome-Wide Patterns of Variation
Jun Z. Li,1,2*{dagger} Devin M. Absher,1,2* Hua Tang,1 Audrey M. Southwick,1,2 Amanda M. Casto,1 Sohini Ramachandran,4 Howard M. Cann,5 Gregory S. Barsh,1,3 Marcus Feldman,4{ddagger} Luigi L. Cavalli-Sforza,1{ddagger} Richard M. Myers1,2{ddagger}

Human genetic diversity is shaped by both demographic and biological factors and has fundamental implications for understanding the genetic basis of diseases. We studied 938 unrelated individuals from 51 populations of the Human Genome Diversity Panel at 650,000 common single-nucleotide polymorphism loci. Individual ancestry and population substructure were detectable with very high resolution. The relationship between haplotype heterozygosity and geography was consistent with the hypothesis of a serial founder effect with a single origin in sub-Saharan Africa. In addition, we observed a pattern of ancestral allele frequency distributions that reflects variation in population dynamics among geographic regions. This data set allows the most comprehensive characterization to date of human genetic variation.