Archive for February, 2007

ssh-keygen usage and R and Bioconductor notes

Wednesday, February 28th, 2007

There is a very good webpage for editing the keys for ssh, which could be very handy if setup this key thing.
I should have known this very long time ago.

There are also very nice R and Bioconductor notes, which were used for workshops came up to me today. It is worthy to recommend (link).

The Crick family home page.

Tuesday, February 27th, 2007

I’m surprised of finging out this Crick family homepage (link). The most intimated name is Francis Crick who shared Nobel Prize with Jim Watson and Perunz Wilkins. The only biography for Francis Crick is “Francis Crick Discoverer of the Genetic Code” by Matt Ridley.

Laplacian of Gaussian and zero crosssing

Monday, February 26th, 2007

I found a website for introductions of Laplacian of Gaussian(LoG), zeor crossing, and edge detection. It is worthy recommendation.
This is part of image processing techniques. Laplacian is used to find the zero points (inflection points) of a signal. Those points could then be used to define the edges (peaks). In the above website there are some nice applications. The Gaussian filter here is to smooth the signal and then convolve with the signal. Since the Laplacian and Gassian are commute in the convolution, so that, in practice, we do the LoG before convolve them with the signal, which make life a little bit easier. Sometimes people use the wavelet filter to denoise the signal before they use LoG to do edge detection.

It snowed again.

Sunday, February 25th, 2007

I have to say something about the weather in Lincoln.

It started with a unusual warm winter. I even couldn’t believe it was winter in the north. Then in the last day of 2006, it snowed, very very heavily. At that moment, I began to taste the winter weather in the north. I have been fully aware of it since then. The snow on the ground melt a little then there were number of snowing. So, the ground have been covered by snow for 90% of the time in 2007.

To make things worse, it snowed very heavily again at Saturday night. I walked to home as a snowman, I mean, literally.

I love snow, but March is coming…

Non-synonymous single nucleotide polymorphisms phenotypic effects prediction

Saturday, February 24th, 2007

Prediction of the phenotypic effects of non-synonymous single nucleotide polymorphisms using structural and evolutionary information.

Lei Bao and Yan Cui, Bioinformatics 21, 2185-2190, 2005

Single nucleotide polymorphisms (SNP) occurs very often so that about 90% sequence variants are different in single nucleotide bases. Some SNPs don’t cause any altered coding sequences, they are named synonymous SNPs. While non-synonymous SNPs (nsSNPs) lead to sequence alteration. There are larger numbers of nsSNPs functioning neutral, but others may cause deleterious effects on protein functions. So, the nsSNPs is a good topic for computational biologists to predict its association with disseases.

In this paper, the authors introduced a representation using structural and evolutionary information for the prediction. The SVM support machine and random forrest (RF) learning were used to implement the training and prediction.  The usage of additional structural information was claimed to contribute to better performance compared to SIFT, a already established algorithm for nsSNPs classification.

The paper is informative. It explained the literature, and reiterated the point from a computational persepective, also pointed out the (dis)advantages of SVM and RF. They stated that SVM is not good for correlated features while RF could overcome it.

But, I should also point out that there is a new “evidence” of the role of synonymous SNPs. A very latest paper found that synonymous SNPs changed the function of the protein in which they occur. It means there is a rare codon marked by this synonymous SNPs.  The paper is :

A “Silent” Polymorphism in the MDR1 Gene Changes Substrate Specificity
Kimchi-Sarfaty, C., et. al., Science 315, 525-528, 2007

Let’s wait and see.

Amphidynamic crystal

Thursday, February 22nd, 2007

There was a presentation by Dr. Miguel A. Garcia-Garibay on amphidynamic crystals. It was a very nice talk. He was trying to develop materials with both crystal and liquid characteristics. The goal was achieved with success, well, of course, more need to be done. It would be very exciting if the material is well developped, at least, one application would be 3D moniter!

The paper was published on PNAS 2005 102: 10771-10776.

Bidirectional promoters

Tuesday, February 20th, 2007

I noticed an interesting fact that in genomes there are some head-to-head genes. Of course, it was discovered by somebody else long time ago. These pairs of genes go to two directions and their 5′ ends are very close, such as, 1k base pairs, so they are called bidirectional genes. As one can imagine, the promoters for those gene pairs might have characteristic impact on gene expressions, i.e., coexpressed, mutually repressed.

One of the papers started the bidirectional promoters is

An Abundance of Bidirectional Promoters in the Human Genome
Genome Research 14:62-66, 2004

From that paper, those promoters are highly likely located in CpG islands. DNA-repair genes are overreprsented in the bidirectional genes, so are chaperone proteins and DEAD-box RNA-helicases.

Quantile normalization

Tuesday, February 20th, 2007

Finally, I decided to take a look at what is the quantile normalization doing, which is widely used in array analysis and introduced by Ben Bolstad. It sorts all the probes according to their measurements for each individual sample. Then assign the average of each row to all its probes in that row. The last step is putting every probe back to where they are before the sorting.
Well done.

As Bolstad stated, the assumption is that the underlying intensities distribution is common across chips. 

 It is worthy pasting here the warnings from the author (pdf):

“One problem with this method is that in the tails in particular, where we might expect greater diferentiation between chips, the normalized values are going to be identical. A modification has been implemented that allows greater differentiation. This works by scaling and centering extreme tail values appropriately without affecting the corresponding quantiles in the other chips. Boxplots can be seen in Fgure 7. Other modifications are also being explored.
Generally speaking one is computing an expression measure for a probeset based upon multiple PM probes or PM/MM probe pairs, thus it may be acceptable to use the uncorrected method.
For smaller numbers of chips, especially when dealing with just 2 chips, a pairwise normalizer may be preferable.”

Year of pig

Sunday, February 18th, 2007

Another new year eve just passed. This time we celebrate the Chinese lunar new year and it is year of pig or ABC news calls it golden pig. When I was young, I believe I’m still yough though, I always wished everyday is the new year. Now, my first reaction is what? new year? again? That is impossible!

Well, you see, yough folks prefer happiness while old ones deny it. Growing up means the process of lacking imagination and sensitivity of joy.

Still so many things pile up on my table and it is not imagination.

Discovery of Histone H3 Lysine 56 Acetyltransferase: Two papers

Saturday, February 17th, 2007

1. Yeast
Rtt109 Promotes Genome Stability by Acetylating Histone H3 on Lysine 56
Science 2 February 2007: Vol. 315. no. 5812, pp. 649 – 652 (Driscoll, R., et. al.)

2. Rtt109 Acetylates Histone H3 Lysine 56 and Functions in DNA Replication

Science 2 February 2007: Vol. 315. no. 5812, pp. 653-655 (Han, J., et. al.)

It is amazing to read two papers showing a same but independent discovery of the novel histone acetyltransferase (HAT). In particular, the two approaches are based on different reasoning. Histone posttranslational modifications are critical in regulating DNA damage response. There are at least three histone modifications which are commonly studied, actylation, phosphorylation, and methylation.

[According to wikipedia, there are also ubiquitination, sumoylation, citrullination, and ADP ribosylation.]

The two papers focus on studying the actylation of the residues Lysine 56 on the tails of Histone H3.
Driscoll approached by starting with an observed fact, “regulation of retrotransposition by Saccharomyces cerevisiae Ty1 transposon is linked to the DNA-damage response(DDR)”. There must be some genes are positively related to the DDR factors. RTT109 was linked to the DDR by other previous study. So, how exactly is RTT109 linked to DDR? Then the authors picked RTT109 and formulated a series of settings to narrow its roles. From those experiments,  it is clear that RTT109 is catalyzing the acetylation of H3K56. One thing needs to be noticed is that Asf1p enhances the activity of the H3 acetylation. The authors then suggested that Asf1p optimize the substrate for H3 acetylation by RTT109.

Han’s approach is different. They started with the fact that H3K56 is transiently acetylated during S phase. This acetylation plays an important role because of its high sensitivity toward certain DNA-damaing agents. The enzymes catalyzing the deacetylation of H3K56 has been discovered, so there ought to be existing a histone H3K56 acetyltransferase. Who is it and how does it work? They then screened 4700 viable mutants to measure their effect on H3K56 acetylation. At last, two candidates, ASF1 and RTT109. The authors then argued that “ASF1 is a histone chaperone and does not appear to have intrinsic HAT activity.” So, they were going to only focus on RTT109. The argument made by the authors is that Rtt109 may use a negatively charged residue to catalyze the process. They then replaced aspartate and glutamate residues with alanines to study its function. D89A and DD287 288AA resulted in the loss of H3K56 acetylation.

Since Rtt109 doesn’t have homology to other known HATs, Rtt109 is a novel acetyltransferase for H3K56 acetylation.