http://www.synapse2disease.ca/en_projet.html
The Synapse to Disease (S2D) project was initiated in 2006, following a successful funding application to Genome Canada and Genome Quebec. The original aim of the S2D project was to identify synaptic genes that cause or predispose an individual to neurodevelopmental diseases such as autism, mental retardation, schizophrenia, and Tourette Syndrome. Contrary to single-gene diseases such as Huntington or Duchenne Muscular Dystrophy, the diseases selected for S2D project result from the interaction of multiple genetic factors (genes) and environmental factors. They are therefore classified as "complex", and represent a major challenge for genetic analysis.
Thursday, April 28, 2011
biomarkers in gene expression
backward selection or forward selection that can select top most discriminating biomarkers
http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5593146&tag=1
http://www.stat.ubc.ca/~rollin/teach/643w04/lec/node42.html
Forward selection has drawbacks, including the fact that each addition of a new variable may render one or more of the already included variables non-significant. An alternate approach which avoids this is backward selection. Under this approach, one starts with fitting a model with all the variables of interest (following the initial screen). Then the least significant variable is dropped, so long as it is not significant at our chosen critical level. We continue by successively re-fitting reduced models and applying the same rule until all remaining variables are statistically significant.
http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5593146&tag=1
http://www.stat.ubc.ca/~rollin/teach/643w04/lec/node42.html
Forward selection has drawbacks, including the fact that each addition of a new variable may render one or more of the already included variables non-significant. An alternate approach which avoids this is backward selection. Under this approach, one starts with fitting a model with all the variables of interest (following the initial screen). Then the least significant variable is dropped, so long as it is not significant at our chosen critical level. We continue by successively re-fitting reduced models and applying the same rule until all remaining variables are statistically significant.
Wednesday, April 27, 2011
Unbalanced data sets in machine learning
http://www.openstarts.units.it/dspace/bitstream/10077/4002/1/Menardi%20Torelli%20DEAMS%20WPS2.pdf
It has been widely reported that the class imbalance heavily compromises the process of learning, because the model tends to focus on the prevalent class and to ignore the rare events (Japkowicz and Stephen, 2002).
However, unless the classes are perfectly separable (Hand and Vinciotti, 2003) or the complexity of the problem is low (Japkowicz and Stephen, 2002),
neglecting the unbalance leads to heavy consequences, both in model estimation and when the evaluation of the accuracy of the estimated model has to be measured.
What typically happens in such a situation is that standard classifiers tend to be overwhelmed by the prevalent class and ignore the rare examples.
Fixes:
1. A first approach to this class of methods produces some modification of the classifier in order to compensate the imbalance. This approach is generally applied to classifiers whose training is based on the optimization of some function related to the overall accuracy.
2. Solutions at the data level for dealing with unbalanced classes basically focus on altering the class distribution in order to get a more balanced sample. (oversampling and undersampling) The reason that altering
the class distribution of the training data aids learning with highly skewed datasets
is that it effectively imposes non-uniform misclassification costs.
It has been widely reported that the class imbalance heavily compromises the process of learning, because the model tends to focus on the prevalent class and to ignore the rare events (Japkowicz and Stephen, 2002).
However, unless the classes are perfectly separable (Hand and Vinciotti, 2003) or the complexity of the problem is low (Japkowicz and Stephen, 2002),
neglecting the unbalance leads to heavy consequences, both in model estimation and when the evaluation of the accuracy of the estimated model has to be measured.
What typically happens in such a situation is that standard classifiers tend to be overwhelmed by the prevalent class and ignore the rare examples.
Fixes:
1. A first approach to this class of methods produces some modification of the classifier in order to compensate the imbalance. This approach is generally applied to classifiers whose training is based on the optimization of some function related to the overall accuracy.
2. Solutions at the data level for dealing with unbalanced classes basically focus on altering the class distribution in order to get a more balanced sample. (oversampling and undersampling) The reason that altering
the class distribution of the training data aids learning with highly skewed datasets
is that it effectively imposes non-uniform misclassification costs.
Tuesday, April 26, 2011
Monday, April 25, 2011
skin cells directly into neural stem cells
http://www.sciencedaily.com/releases/2011/04/110425153600.htm?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+sciencedaily+%28ScienceDaily%3A+Latest+Science+News%29
Dr. Ding focuses on reprogramming skin cells into neural stem cells using the existing iPS technology -- but with a twist. Dr. Ding never lets the cells enter the pluripotent state of iPS cells, in which they could develop into any type of cell. Instead he uses yet another cocktail of factors to transform the skin cells directly into neural stem cells. Avoiding the pluripotent state is important because it avoids the potential danger that "rogue" iPS cells could develop into a tumor if used to replace or repair damaged organs or tissue.
Dr. Ding focuses on reprogramming skin cells into neural stem cells using the existing iPS technology -- but with a twist. Dr. Ding never lets the cells enter the pluripotent state of iPS cells, in which they could develop into any type of cell. Instead he uses yet another cocktail of factors to transform the skin cells directly into neural stem cells. Avoiding the pluripotent state is important because it avoids the potential danger that "rogue" iPS cells could develop into a tumor if used to replace or repair damaged organs or tissue.
Amazon seller lists book at $23,698,655.93 -- plus shipping
http://www.cnn.com/2011/TECH/web/04/25/amazon.price.algorithm/index.html?hpt=C2
The incident highlights a little-known fact about e-commerce sites such as Amazon: Often, people don't create and update prices; computer algorithms do.
The incident highlights a little-known fact about e-commerce sites such as Amazon: Often, people don't create and update prices; computer algorithms do.
Friday, April 22, 2011
Bioinformatics visualization
http://www.sfu.ca/~shaw/
http://biov.iat.sfu.ca/IMASProjectWeb
BrainFrame-VDA09.pdf (282K)
BrainFrame: A Knowledge Visualization System for the Neurosciences,
Steven J. Barnes, Chris D. Shaw
Proceedings of VDA 2009 Conference on Visualization and Data Analysis 2009 , San Jose, California, January 19-22, 2009, pp. 72430F.1-10 (10 pages).
http://biov.iat.sfu.ca/IMASProjectWeb
BrainFrame-VDA09.pdf (282K)
BrainFrame: A Knowledge Visualization System for the Neurosciences,
Steven J. Barnes, Chris D. Shaw
Proceedings of VDA 2009 Conference on Visualization and Data Analysis 2009 , San Jose, California, January 19-22, 2009, pp. 72430F.1-10 (10 pages).
Thursday, April 21, 2011
What is a PhD really worth?
http://www.nature.com/naturejobs/2011/110421/full/nj7343-381a.html?WT.ec_id=NATUREjobs-20110421
http://graduate-school.phds.org/
http://sites.nationalacademies.org/pga/resdoc/index.htm
Frugal science graduate students emerge with a very important life lesson: money does not buy happiness (although it certainly can make misery a lot more comfortable), and living frugally is better than amassing tens of thousands of dollars of debt.
But I believe the most important lesson is that no programme of higher education can guarantee its graduates gainful and lucrative employment. At best, a graduate programme in any discipline can provide its students with key skills, knowledge and abilities. How the graduates apply that learning is up to them.
Focused seminars in areas such as communication, business basics and public policy would go a long way towards strengthening the capabilities of PhD students and improving their career prospects.
http://graduate-school.phds.org/
http://sites.nationalacademies.org/pga/resdoc/index.htm
Frugal science graduate students emerge with a very important life lesson: money does not buy happiness (although it certainly can make misery a lot more comfortable), and living frugally is better than amassing tens of thousands of dollars of debt.
But I believe the most important lesson is that no programme of higher education can guarantee its graduates gainful and lucrative employment. At best, a graduate programme in any discipline can provide its students with key skills, knowledge and abilities. How the graduates apply that learning is up to them.
Focused seminars in areas such as communication, business basics and public policy would go a long way towards strengthening the capabilities of PhD students and improving their career prospects.
Wednesday, April 20, 2011
heatmap using a custom distance for hclust clustering
# Correlation distance function.
cordist <- function(x ) {
return( as.dist(( 1 - cor(t(x), use = "pairwise.complete.obs"))))
}
# Getting nice aligned clustering trees. Assume given my.dists.
clus<-hclust(my.dists,method = "a")
# Plot heatmap
#heatmap(x, Rowv=as.dendrogram(hclust(cordist(x))))
heatmap(x, Rowv=as.dendrogram(clus))
cordist <- function(x ) {
return( as.dist(( 1 - cor(t(x), use = "pairwise.complete.obs"))))
}
# Getting nice aligned clustering trees. Assume given my.dists.
clus<-hclust(my.dists,method = "a")
# Plot heatmap
#heatmap(x, Rowv=as.dendrogram(hclust(cordist(x))))
heatmap(x, Rowv=as.dendrogram(clus))
Einstein - Imagination is more important than knowledge.
As Einstein so eloquently put it, "Imagination is more important than knowledge." And that creativity, that imagination, that ability to find that new approach to the problem everyone else is working on is what's going to make you a great scientist.
Imagination is more important than knowledge.
One of the main thing to learn in graduate studies is how to think critically, synthesize, create a new process, not simply follow (tech work) which is tech dependent that may become obsolete in the future.
http://scienceblogs.com/startswithabang/2010/08/advice_for_young_aspiring_scie.php
Imagination is more important than knowledge.
One of the main thing to learn in graduate studies is how to think critically, synthesize, create a new process, not simply follow (tech work) which is tech dependent that may become obsolete in the future.
http://scienceblogs.com/startswithabang/2010/08/advice_for_young_aspiring_scie.php
What happens when the spotlight shines on the young scientists?
http://www.nature.com/nature/journal/v467/n7317_supp/full/467S22a.html
Box 1: Advice to laureates
* Reply to emails from students within twelve hours
* Don't dictate a student's life
* Give creative freedom
* Foster relationship among students in the lab, not just with them
* Let students develop their 'voice' when writing papers
* Communicate your science to the public by using the media
Box 2: Advice to students
* Choose a supervisor who does not travel too much
* Don't try to please your supervisor all the time, be prepared to challenge them
* Put questions to your supervisor, but think of some possible suggestions beforehand
* Assume your supervisor is wrong and develop your own way to approach the problem
* Idealism regarding science in politics is good, but be aware that it will be a steep challenge
* Don't give up too easily
Box 1: Advice to laureates
* Reply to emails from students within twelve hours
* Don't dictate a student's life
* Give creative freedom
* Foster relationship among students in the lab, not just with them
* Let students develop their 'voice' when writing papers
* Communicate your science to the public by using the media
Box 2: Advice to students
* Choose a supervisor who does not travel too much
* Don't try to please your supervisor all the time, be prepared to challenge them
* Put questions to your supervisor, but think of some possible suggestions beforehand
* Assume your supervisor is wrong and develop your own way to approach the problem
* Idealism regarding science in politics is good, but be aware that it will be a steep challenge
* Don't give up too easily
Tuesday, April 19, 2011
Proteomics and forestry - Jörg Bohlmann
Jörg Bohlmann
Secondary metabolism; Plant defense against insects and insect-associated pathogens; Forestry and grapevine genomics
http://www.msl.ubc.ca/faculty/bohlmann
Secondary metabolism; Plant defense against insects and insect-associated pathogens; Forestry and grapevine genomics
http://www.msl.ubc.ca/faculty/bohlmann
DebugMode - Wink = flash tutorial and presentation creation software
http://www.debugmode.com/wink/
Wink is a Tutorial and Presentation creation software, primarily aimed at creating tutorials on how to use software (like a tutor for MS-Word/Excel etc). Using Wink you can capture screenshots, add explanations boxes, buttons, titles etc and generate a highly effective tutorial for your users.
Wink is a Tutorial and Presentation creation software, primarily aimed at creating tutorials on how to use software (like a tutor for MS-Word/Excel etc). Using Wink you can capture screenshots, add explanations boxes, buttons, titles etc and generate a highly effective tutorial for your users.
Dalian, China - DNA Day
http://www.dnaday.com/
Dalian-China’s Best Tourist City. With beautiful scenery, a nice climate and fast development, this romantic city ranks among China’s best tourist destinations in many people’s minds. It is also a trading and financial center in northeastern Asia and has gained the name ‘Hong Kong of Northern China.’
Dalian-China’s Best Tourist City. With beautiful scenery, a nice climate and fast development, this romantic city ranks among China’s best tourist destinations in many people’s minds. It is also a trading and financial center in northeastern Asia and has gained the name ‘Hong Kong of Northern China.’
Glial
http://med.stanford.edu/ism/2009/september/glia-0921.html
Alzheimer’s disease is characterized by massive synapse loss.
As one ascends the scale of evolutionary complexity, an increasing proportion of the brain’s cells are glial. In the simple nematode worm, they’re sparse; in a fruit fly, they’re up to 25 percent; in a mouse, about 65 percent. In a human brain, behind every great neuron stand nine great glial cells.
http://med.stanford.edu/ism/images/featureStories/glia-illo-092109.jpg
There are three main types of glial cells. Oligodendrocytes (1) send projections that wrap axons (2) – long, signal-carrying portions of neurons (3) – in sheathes of a fatty substance called myelin (4), speeding signal conduction. Microglia (5) are, essentially, the brain’s immune cells, but they also monitor neighboring brain cells for damage and gobble up debris, and they probably have other functions, too. Astrocytes (6) carry on a host of activities. Their long extensions can monitor levels of neuronal activity either along axons at synapses (7) – junctions that relay signals from one neuron to the next – and, when those activity levels are high, signal to local blood vessels (8) to dilate, increasing blood supply to hard-working neurons. Astrocytes also produce and secrete substances that have a major influence on the formation and elimination of synapses.
Alzheimer’s disease is characterized by massive synapse loss.
As one ascends the scale of evolutionary complexity, an increasing proportion of the brain’s cells are glial. In the simple nematode worm, they’re sparse; in a fruit fly, they’re up to 25 percent; in a mouse, about 65 percent. In a human brain, behind every great neuron stand nine great glial cells.
http://med.stanford.edu/ism/images/featureStories/glia-illo-092109.jpg
ALS vs MS
http://www.cnsonline.org/www/archive/ms/ms-04.html
"sclerosis," which literally means hardening (as a result of increased connective tissue or glia).
Multiple sclerosis is a disease of myelin, not primarily of nerve cells. This myelin surrounds the axons, or the long process of the nerve cell.
The principle characteristic in the pathology of amyotrophic lateral sclerosis(ALS) is loss of motor nerve cells in the anterior horns of the spinal cord and in the motor nuclei of the brain stem.
Thus, it is not primary demyelination, as it is in multiple sclerosis, that is the primary destructive effect in ALS.
"sclerosis," which literally means hardening (as a result of increased connective tissue or glia).
Multiple sclerosis is a disease of myelin, not primarily of nerve cells. This myelin surrounds the axons, or the long process of the nerve cell.
The principle characteristic in the pathology of amyotrophic lateral sclerosis(ALS) is loss of motor nerve cells in the anterior horns of the spinal cord and in the motor nuclei of the brain stem.
Thus, it is not primary demyelination, as it is in multiple sclerosis, that is the primary destructive effect in ALS.
Monday, April 18, 2011
Mantel test
The Mantel test, named after Nathan Mantel, is a statistical test of the correlation between two matrices.
The test is commonly used in ecology, where the data are usually estimates of the "distance" between objects such as species of organisms. For example, one matrix might contain estimates of the genetic distances (i.e., the amount of difference between two different genomes) between all possible pairs of species in the study, obtained by the methods of molecular systematics; while the other might contain estimates of the geographical distance between the ranges of each species and every other species.
http://en.wikipedia.org/wiki/Mantel_test
The test is commonly used in ecology, where the data are usually estimates of the "distance" between objects such as species of organisms. For example, one matrix might contain estimates of the genetic distances (i.e., the amount of difference between two different genomes) between all possible pairs of species in the study, obtained by the methods of molecular systematics; while the other might contain estimates of the geographical distance between the ranges of each species and every other species.
http://en.wikipedia.org/wiki/Mantel_test
Condtional Random Field - CRF (Machine Learning)
http://en.wikipedia.org/wiki/Conditional_random_field
A conditional random field (CRF) is a type of discriminative undirected probabilistic graphical model. It is most often used for labeling or parsing of sequential data, such as natural language text or biological sequences[1] and computer vision[2] . Specifically, CRFs find applications in shallow parsing[3] , named entity recognition[4] and gene finding, among other tasks, being an alternative to the related hidden Markov models.
http://www.inference.phy.cam.ac.uk/hmw26/crf/
The primary advantage of CRFs over hidden Markov
models is their conditional nature, resulting in the relaxation of the indepen-
dence assumptions required by HMMs in order to ensure tractable inference.
CRFs outperform both MEMMs
and HMMs on a number of real-world sequence labeling tasks
Nando de Freitas
http://www.cs.ubc.ca/~nando/
http://www-stat.stanford.edu/~tibs/ElemStatLearn/download.html
David JC MacKay
http://www.inference.phy.cam.ac.uk/mackay/itila/
A conditional random field (CRF) is a type of discriminative undirected probabilistic graphical model. It is most often used for labeling or parsing of sequential data, such as natural language text or biological sequences[1] and computer vision[2] . Specifically, CRFs find applications in shallow parsing[3] , named entity recognition[4] and gene finding, among other tasks, being an alternative to the related hidden Markov models.
http://www.inference.phy.cam.ac.uk/hmw26/crf/
The primary advantage of CRFs over hidden Markov
models is their conditional nature, resulting in the relaxation of the indepen-
dence assumptions required by HMMs in order to ensure tractable inference.
CRFs outperform both MEMMs
and HMMs on a number of real-world sequence labeling tasks
Nando de Freitas
http://www.cs.ubc.ca/~nando/
http://www-stat.stanford.edu/~tibs/ElemStatLearn/download.html
- Deep Learning Website.
- Geoff Hinton's Website.
- Yann Lecun's Website.
- Andrew Ng's Website.
- Jason Weston's Website.
- Russ Salakhutdinov's Website.
- The book of Kevin Murphy.
- Hastie, Tibshirani and Friedman: The elements of statistical learning.
- Machine learning video lectures
- Why stats: NYTimes article
- The following handout should help you with linear algebra revision: PDF
David JC MacKay
http://www.inference.phy.cam.ac.uk/mackay/itila/
QR - quick response code
A QR code (short for Quick Response) is a specific matrix barcode (or two-dimensional code), readable by dedicated QR barcode readers and camera phones. The code consists of black modules arranged in a square pattern on a white background. The information encoded can be text, URL or other data.
http://en.wikipedia.org/wiki/QR_code
http://www.qrinkle.com/
http://goo.gl/
http://r20.rs6.net/tn.jsp?llr=lgkxgtbab&et=1106584988522&s=6570&e=001x4WpKVoIrGnfLd9yneahKEE3mjYjdFZIwTcIHFgRPYs_6bCdRAFiR8PUCmm9QmxGRLWBe3jjEC_VfcvnieKl3NxtV9EMNTwG_870hXx2cybexx0G0I6AVH_VGw4f4ZGz5GzXUlQg_dohlELdczK8KAJ6Hk6-jYppREdiikufjSq6TuCe_Cf48Fy7KalBZuRSozJRgFCPJ5lc-sKO-Cs60_8ARJUtaOnMd5opAhZmewAAvhIHYkZ78SvEZNIRRlfR_PHt-7nQpLY=
http://en.wikipedia.org/wiki/QR_code
http://www.qrinkle.com/
http://goo.gl/
http://r20.rs6.net/tn.jsp?llr=lgkxgtbab&et=1106584988522&s=6570&e=001x4WpKVoIrGnfLd9yneahKEE3mjYjdFZIwTcIHFgRPYs_6bCdRAFiR8PUCmm9QmxGRLWBe3jjEC_VfcvnieKl3NxtV9EMNTwG_870hXx2cybexx0G0I6AVH_VGw4f4ZGz5GzXUlQg_dohlELdczK8KAJ6Hk6-jYppREdiikufjSq6TuCe_Cf48Fy7KalBZuRSozJRgFCPJ5lc-sKO-Cs60_8ARJUtaOnMd5opAhZmewAAvhIHYkZ78SvEZNIRRlfR_PHt-7nQpLY=
Sunday, April 17, 2011
Friday, April 15, 2011
VanBUG Career Panel Talk
- if you like what you're doing now, great, otherwise move on
- interviews are made to measure your stress level
- PhD can sometimes close more doors than it opens (might need to relocate ...)
- Some companies allow you to publish while working!
- Network network network and did I say network?
- Conference blogging as a career, blog in general, (caveat: might have setbacks, eg. bad comments, so try to be objective, exclude personal life)
- get some web presence, work on some open-source projects (it shows that you can work with others)
- There's no 'Bioinformatics' job title in the Gov't search, pick 'Scientist' or computer scientist or something close ...
- Follow your passion, side-projects might blossom
- Know where you want to go and work towards it, eg. PhD requires you to drive your own research
- ~98% of startups fail, need to find the right people to work with, those people with complementary skills (entrepreneurial)
- Great researchers have ideas, are able to see through it to the end and excellent collaborators
Thursday, April 14, 2011
S. Johnson - Bitter Love
Love is the wisdom of the fool and the folly of the wise.Love is the wisdom of the fool and the folly of the wise.
Metric (TFM) file not found
I tried to use \usepackage{times} to set my document in the times font, I get an error message: "! Font OT1/ptm/m/n/10=ptmr7t at 10.0pt not loadable: Metric (TFM) file not found" Why doesn't it work?
install packages
texlive-fonts-recommended
latex-xtf-fonts
ttf-symbol-replacement
install packages
texlive-fonts-recommended
latex-xtf-fonts
ttf-symbol-replacement
Tuesday, April 12, 2011
heatmap.plus - for plotting multiple RowSideColors and ColSideColors
http://cran.r-project.org/web/packages/heatmap.plus/index.html
It seems that it can't plot more Col/RowSideColors if there's only one row/column, work around is to duplicate
ColSideColors <- t(as.matrix(rbind(col.a, col.a)))
It seems that it can't plot more Col/RowSideColors if there's only one row/column, work around is to duplicate
ColSideColors <- t(as.matrix(rbind(col.a, col.a)))
Monday, April 11, 2011
LaTex special characters
http://web.science.mq.edu.au/~rdale/resources/writingnotes/latexstyle.html#dashes
200 - 500 bp is written as "200 -- 500 bp"
http://latex.knobs-dials.com/
200 - 500 bp is written as "200 -- 500 bp"
tilde ~0.5 use $\sim$0.5http://theoval.cmp.uea.ac.uk/~nlct/latex/novices/symbols.html#sec:chars http://tex.stackexchange.com/questions/9363/how-does-one-insert-a-backslash-or-a-tilde-into-latexhttp://latex.knobs-dials.com/
Can't get root shell when fsck check fails in ubuntu
Try editing your regular boot entry in grub (e shortcut and appending
http://superuser.com/questions/215590/how-to-boot-grub2-into-the-simplest-linux-shell
$ sudo gedit /boot/grub/grub.cfg
menuentry 'Ubuntu recovery with Ctrl+X then fsck /dev/sda4 (home)' --class ubuntu --class gnu-linux --class gnu --class os {
recordfail
insmod ext2
set root='(hd0,3)'
search --no-floppy --fs-uuid --set dc61149c-085c-4b71-a92f-127a6b831719
echo 'Loading Linux 2.6.32-25-generic ...'
linux /boot/vmlinuz-2.6.32-25-generic root=UUID=dc61149c-085c-4b71-a92f-127a6b831719 ro single init=/bin/sh
echo 'Loading initial ramdisk ...'
initrd /boot/initrd.img-2.6.32-25-generic
}
https://help.ubuntu.com/community/Grub2
When all else fails, try booting to the Windows partition, with a boot cd / usb
init=/bin/sh to end of the line beginning with linux .... After you will boot it (Ctrl-x), you will get a root shell immediately.http://superuser.com/questions/215590/how-to-boot-grub2-into-the-simplest-linux-shell
$ sudo gedit /boot/grub/grub.cfg
menuentry 'Ubuntu recovery with Ctrl+X then fsck /dev/sda4 (home)' --class ubuntu --class gnu-linux --class gnu --class os {
recordfail
insmod ext2
set root='(hd0,3)'
search --no-floppy --fs-uuid --set dc61149c-085c-4b71-a92f-127a6b831719
echo 'Loading Linux 2.6.32-25-generic ...'
linux /boot/vmlinuz-2.6.32-25-generic root=UUID=dc61149c-085c-4b71-a92f-127a6b831719 ro single init=/bin/sh
echo 'Loading initial ramdisk ...'
initrd /boot/initrd.img-2.6.32-25-generic
}
https://help.ubuntu.com/community/Grub2
When all else fails, try booting to the Windows partition, with a boot cd / usb
underconnectivity theory of autism
Inter-regional brain communication and its disturbance in autism
http://www.frontiersin.org/systems_neuroscience/10.3389/fnsys.2011.00010/full
underconnectivity theory of autism postulates that individuals with autism have a reduced communication bandwidth between frontal and posterior cortical areas, which constrains the psychological processes that rely on the integrated functioning of frontal and posterior brain networks.
Thus, brain volume measurements have revealed that the rate of brain growth in autism slows after age 4, leading to a decreased volume of white matter in adolescents with autism relative to neurotypical adolescents. Given that white matter is the medium which is used for inter-regional brain communication, it seems incontrovertible that brain connectivity is disrupted in autism.
Altered functional connectivity has also been found in other disorders, including schizophrenia (Meyer-Lindenberg et al., 2001), attention deficit hyperactivity disorder (Tian et al., 2006), multiple sclerosis (Au Duong et al., 2005), and dyslexia (Pugh et al., 2000). These findings suggest that disordered brain connectivity may underlie a variety of cognitive impairments.
While autism is primarily associated with frontal–posterior underconnectivity, preliminary evidence suggests that these other disorders are linked with impairments in other types of connections (Pugh et al., 2000; Meyer-Lindenberg et al., 2001; Au Duong et al., 2005; Tian et al., 2006).
Recent findings of atypical patterns in both functional and anatomical connectivity in autism have established that autism is a not a localized neurological disorder, but one that affects many parts of the brain in many types of thinking tasks. fMRI studies repeatedly find evidence of decreased coordination between frontal and posterior brain regions in autism, as measured by functional connectivity.
Sunday, April 10, 2011
Epigenetics: Impact of DNA methylation
http://ows.molgen.mpg.de/2009/lectures/steinhoff.pdf
Bioinformation. 2010 Jan 23;4(7):331-7.
Computational Epigenetics: the new scientific paradigm.
Lim SJ, Tan TW, Tong JC.
http://www.ncbi.nlm.nih.gov/pubmed/20978607
Computational epigenetics
http://bioinformatics.oxfordjournals.org/content/24/1/1.abstract
http://www.youtube.com/watch?v=eYrQ0EhVCYA
Nat Biotechnol. 2010 Oct;28(10):1045-8.
The NIH Roadmap Epigenomics Mapping Consortium.
Bernstein BE, Stamatoyannopoulos JA, Costello JF, Ren B, Milosavljevic A, Meissner A, Kellis M, Marra MA, Beaudet AL, Ecker JR, Farnham PJ, Hirst M, Lander ES, Mikkelsen TS, Thomson JA.
Broad Institute of Harvard and MIT, Cambridge, Massachusetts, USA. Bernstein.Bradley@mgh.harvard.edu
http://www.ncbi.nlm.nih.gov/pubmed/20944595
http://commonfund.nih.gov/epigenomics/
Bioinformation. 2010 Jan 23;4(7):331-7.
Computational Epigenetics: the new scientific paradigm.
Lim SJ, Tan TW, Tong JC.
http://www.ncbi.nlm.nih.gov/pubmed/20978607
Computational epigenetics
http://bioinformatics.oxfordjournals.org/content/24/1/1.abstract
http://www.youtube.com/watch?v=eYrQ0EhVCYA
Nat Biotechnol. 2010 Oct;28(10):1045-8.
The NIH Roadmap Epigenomics Mapping Consortium.
Bernstein BE, Stamatoyannopoulos JA, Costello JF, Ren B, Milosavljevic A, Meissner A, Kellis M, Marra MA, Beaudet AL, Ecker JR, Farnham PJ, Hirst M, Lander ES, Mikkelsen TS, Thomson JA.
Broad Institute of Harvard and MIT, Cambridge, Massachusetts, USA. Bernstein.Bradley@mgh.harvard.edu
http://www.ncbi.nlm.nih.gov/pubmed/20944595
http://commonfund.nih.gov/epigenomics/
Saturday, April 9, 2011
worry is wasteful - JEWEL 'Hands'
And not to worry 'cause worry is wasteful
And useless in times like these
In the end only kindness matters
http://www.azlyrics.com/lyrics/jewel/hands.html
And useless in times like these
In the end only kindness matters
http://www.azlyrics.com/lyrics/jewel/hands.html
DREAM - Towards a Rigorous Assessment of Systems Biology Models: The DREAM3 Challenges
http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0009202
Competition inspired by Critical Assessment of techniques for protein Structure Prediction (CASP)
Competition inspired by Critical Assessment of techniques for protein Structure Prediction (CASP)
Friday, April 8, 2011
Why French Fries Are Such Good Comfort Food
http://healthland.time.com/2011/04/07/why-french-fries-are-good-comfort-food/?hpt=C2
Oxytocin is crucial to the processes that allow love and social contact to reduce stress.
Oxytocin is crucial to the processes that allow love and social contact to reduce stress.
Mark Victor Hansen - Ideas
Ideas attract money, time, talents, skills, energy and other complementary ideas that will bring them into reality.
Thursday, April 7, 2011
Stouffer's method - weighted z-scores
http://en.wikipedia.org/wiki/Fisher%27s_method#Relation_to_Stouffer.27s_Z-score_method
A closely related approach to Fisher's method is based on Z-scores rather than p-values.
One advantage of the Z-score approach is that it is straightforward to introduce weights.
Stouffer, S.A.; E.A. Suchman, L.C. DeVinney, S.A. Star, R.M. Jr. Williams (1949). The American Soldier, Vol.1: Adjustment during Army Life. Princeton University Press, Princeton,.
A closely related approach to Fisher's method is based on Z-scores rather than p-values.
One advantage of the Z-score approach is that it is straightforward to introduce weights.
Stouffer, S.A.; E.A. Suchman, L.C. DeVinney, S.A. Star, R.M. Jr. Williams (1949). The American Soldier, Vol.1: Adjustment during Army Life. Princeton University Press, Princeton,.
Crafting Your Funding Application
www.grad.ubc.ca/awards
www.scholarshipscanada.com
www.studentawards.com
like writing a cover letter
summary
- catchy first paragraph
- relevant
- don't use short form in title
- don't make reviewer feel dumb
- write for the general audience
- define acronyms
- cite reference
- be specific, put numbers, examples, how you would approach the problem
- get help from supervisor about writing it
- be meticulous, ask a friend to proof-read, read it backwards
CIHR
www.cihr-crsh.gc.ca/33043.html#annex1
Grantsmanship:
www.hfsp.org/how/ArtOfGrants.htm
TIMELINE
aug. 1 - colleage review
aug. 12 - supervisor first review
aug. 22 - editor review
sept. 1 - supervisor review
sept. 19 - signatures
sept. 23 - dept. deadline
oct. 7 - deadline
supporting evidence for reference letters
- critical thinking, independence, perseverance, originality, judgement, research ability
www.scholarshipscanada.com
www.studentawards.com
like writing a cover letter
summary
- convince them that your research is important
- summary page is the most critical section of your application
- good structure with headings (background, hypothesis, objectives, significance, method, problem
- catchy first paragraph
- relevant
- don't use short form in title
- don't make reviewer feel dumb
- write for the general audience
- define acronyms
- cite reference
- be specific, put numbers, examples, how you would approach the problem
- get help from supervisor about writing it
- be meticulous, ask a friend to proof-read, read it backwards
CIHR
www.cihr-crsh.gc.ca/33043.html#annex1
Grantsmanship:
www.hfsp.org/how/ArtOfGrants.htm
TIMELINE
aug. 1 - colleage review
aug. 12 - supervisor first review
aug. 22 - editor review
sept. 1 - supervisor review
sept. 19 - signatures
sept. 23 - dept. deadline
oct. 7 - deadline
supporting evidence for reference letters
- critical thinking, independence, perseverance, originality, judgement, research ability
Tuesday, April 5, 2011
R ROC, AUC
source("http://bioconductor.org/getBioC.R")
getBioC("ROCR")
getBioC("GEOquery")
install.packages("e1071")
http://rocr.bioinf.mpi-sb.mpg.de/
data(ROCR.simple) pred <- ROCR::prediction(ROCR.simple$predictions, ROCR.simple$labels)
perf <- performance(pred,"tpr","fpr")
plot(perf,colorize=TRUE)
plot(performance(prediction(c(0,1,-1), labels=c("a","b","a")), "tpr", "fpr"))
performance(prediction(c(0,1,1), labels=c("a","b","a")), "auc")@y.values
getBioC("ROCR")
getBioC("GEOquery")
install.packages("e1071")
http://rocr.bioinf.mpi-sb.mpg.de/
data(ROCR.simple) pred <- ROCR::prediction(ROCR.simple$predictions, ROCR.simple$labels)
perf <- performance(pred,"tpr","fpr")
plot(perf,colorize=TRUE)
plot(performance(prediction(c(0,1,-1), labels=c("a","b","a")), "tpr", "fpr"))
performance(prediction(c(0,1,1), labels=c("a","b","a")), "auc")@y.values
Short-read aligners / mappers
maps to a reference genome, or another somatic genome (tumor vs normal sample) - useful for copy number variations (CNV)
- MAQ
- BowTie, BWA, Soap2 (uses Burrows–Wheeler Transform [38], a
technique previously used for compression) The
BWT string is built by sorting all of the circular
shifts of a string, and concatenating the last characters
of each circular shift 'last-first property'
- mrFAST
pair reads are useful for large structural variations (insertions, deletions, translocations), problem with inserts that are larger than the expected distance between pairs (try hanging-reads - only of the reads are mapped)
ABI Solid's colorspace is useful for SNP analysis
This approach also
demonstrates a key advantage of the color-space
encoding. When one compares a regular, letter-
space read to a known DNA sequence, it is difficult
to determine if a discrepancy is due to a true differ-
ence between the two genomes, or to a sequencing
error. In color-space, we can usually separate the two
explanations: if the difference is due to an SNP
between the genomes, this will lead to two adjacent
color-space changes, as both of the colors that inter-
rogated the nucleotide will change.
On the other hand, a sequencing error will only
affect one color, and therefore can be differentiated
from a SNP.
color-space Smith-Waterman - aligns the reference of all 4 possible translations (like transcript to protein)
problem with repeats, non-uniform coverage
- MAQ
- BowTie, BWA, Soap2 (uses Burrows–Wheeler Transform [38], a
technique previously used for compression) The
BWT string is built by sorting all of the circular
shifts of a string, and concatenating the last characters
of each circular shift 'last-first property'
- mrFAST
pair reads are useful for large structural variations (insertions, deletions, translocations), problem with inserts that are larger than the expected distance between pairs (try hanging-reads - only of the reads are mapped)
ABI Solid's colorspace is useful for SNP analysis
This approach also
demonstrates a key advantage of the color-space
encoding. When one compares a regular, letter-
space read to a known DNA sequence, it is difficult
to determine if a discrepancy is due to a true differ-
ence between the two genomes, or to a sequencing
error. In color-space, we can usually separate the two
explanations: if the difference is due to an SNP
between the genomes, this will lead to two adjacent
color-space changes, as both of the colors that inter-
rogated the nucleotide will change.
On the other hand, a sequencing error will only
affect one color, and therefore can be differentiated
from a SNP.
color-space Smith-Waterman - aligns the reference of all 4 possible translations (like transcript to protein)
problem with repeats, non-uniform coverage
High throughput sequencing
A.V. Dalca, M. Brudno, Briefings in Bioinformatics (2010) 11(1):3-14.
While
GenomeMapper [57] is the first tool to allow for
the simultaneous mapping of HTS reads to multiple
genomes, identifying variants—both SNPs and
SVs—based on many low-coverage individuals is
another important research area, and one which
may prove key to enabling the $1000 genome and
the full promise of personal genomics.
An alternate method for copy-number variation
(CNV) discovery relies on the ‘depth-of-coverage’
(DOC) signal. If a certain genomic region is present
multiple times in the donor genome, more reads will
likely be generated from it, and consequently the
corresponding region in the reference will have
higher coverage (Figure 4D). -- this assumes uniform coverage though!
While SNPs and small indels can be located by
analyzing the mappings of unpaired reads, the iden-
tification of structural variants (SVs), where the
genome is drastically altered, is more difficult with
short reads. For example, a large deletion in the
donor’s genome (i.e. a segment of the reference
not present in the donor) may create split-reads
that cover the location of the deletion (the break-
point), and map to the reference with their two
halves on opposite sides of the deleted segment.
Accordingly, the discovery
of SVs in a genome is typically based on pair-end
sequencing approaches [19]. The two reads are
mapped to the reference genome, with the distance
between them referred to as ‘mapped distance’. This
mapped distance and the relative orientations of the
mapping are then compared to the expected insert
size: if the distance is similar and the orientations are
unchanged, the matepair is termed ‘concordant’, and
is thought to be unlikely to overlap an SV. If, on the
other hand, one of these is different or changed (the
mate pair is called ‘discordant’), it likely overlaps a
variant, such as an insertion (the mapped distance
will be smaller than expected insert size), deletion
(it will be larger) or inversion (the orientation of
one of the two mappings will be opposite from the
expected).
Methods for SV detection with mate pairs can
identify many, but not all SVs. For example,
insertions (in the donor) larger than the insert size
cannot be discovered by these methods, as no mate-
pair will completely span the insertion event.
While
GenomeMapper [57] is the first tool to allow for
the simultaneous mapping of HTS reads to multiple
genomes, identifying variants—both SNPs and
SVs—based on many low-coverage individuals is
another important research area, and one which
may prove key to enabling the $1000 genome and
the full promise of personal genomics.
An alternate method for copy-number variation
(CNV) discovery relies on the ‘depth-of-coverage’
(DOC) signal. If a certain genomic region is present
multiple times in the donor genome, more reads will
likely be generated from it, and consequently the
corresponding region in the reference will have
higher coverage (Figure 4D). -- this assumes uniform coverage though!
While SNPs and small indels can be located by
analyzing the mappings of unpaired reads, the iden-
tification of structural variants (SVs), where the
genome is drastically altered, is more difficult with
short reads. For example, a large deletion in the
donor’s genome (i.e. a segment of the reference
not present in the donor) may create split-reads
that cover the location of the deletion (the break-
point), and map to the reference with their two
halves on opposite sides of the deleted segment.
Accordingly, the discovery
of SVs in a genome is typically based on pair-end
sequencing approaches [19]. The two reads are
mapped to the reference genome, with the distance
between them referred to as ‘mapped distance’. This
mapped distance and the relative orientations of the
mapping are then compared to the expected insert
size: if the distance is similar and the orientations are
unchanged, the matepair is termed ‘concordant’, and
is thought to be unlikely to overlap an SV. If, on the
other hand, one of these is different or changed (the
mate pair is called ‘discordant’), it likely overlaps a
variant, such as an insertion (the mapped distance
will be smaller than expected insert size), deletion
(it will be larger) or inversion (the orientation of
one of the two mappings will be opposite from the
expected).
Methods for SV detection with mate pairs can
identify many, but not all SVs. For example,
insertions (in the donor) larger than the insert size
cannot be discovered by these methods, as no mate-
pair will completely span the insertion event.
Voltaire - "Judge a man by his questions rather than his answers."
"Judge a man by his questions rather than his answers."
Voltaire
Voltaire
Monday, April 4, 2011
What makes clinical research ethical?
http://www.ncbi.nlm.nih.gov/pubmed/10819955
JAMA. 2000 May 24-31;283(20):2701-11.
What makes clinical research ethical?
Warren G. Magnuson Clinical Center, Bldg 10, Room 1C118, National Institutes of Health, Bethesda, MD 20892-1156, USA.
Mechanisms of B-cell lymphoma pathogenesis
1. Ralf Kuppers, “Mechanisms of B-cell lymphoma pathogenesis,” Nat Rev Cancer 5, no. 4 (April 2005): 251-262.
http://www.nature.com/nrc/journal/v5/n4/abs/nrc1589.html
http://www.nature.com/nrc/journal/v5/n4/abs/nrc1589.html
DAVID - Gene name batch viewer
http://david.abcc.ncifcrf.gov/home.jsp
The Database for Annotation, Visualization and Integrated Discovery (DAVID ) v6.7 is an update to the sixth version of our original web-accessible programs. DAVID now provides a comprehensive set of functional annotation tools for investigators to understand biological meaning behind large list of genes. For any given gene list, DAVID tools are able to:
Identify enriched biological themes, particularly GO terms
Discover enriched functional-related gene groups
Cluster redundant annotation terms
Visualize genes on BioCarta & KEGG pathway maps
Display related many-genes-to-many-terms on 2-D view.
Search for other functionally related genes not in the list
List interacting proteins
Explore gene names in batch
Link gene-disease associations
Highlight protein functional domains and motifs
Redirect to related literatures
Convert gene identifiers from one type to another.
And more
The Database for Annotation, Visualization and Integrated Discovery (DAVID ) v6.7 is an update to the sixth version of our original web-accessible programs. DAVID now provides a comprehensive set of functional annotation tools for investigators to understand biological meaning behind large list of genes. For any given gene list, DAVID tools are able to:
Identify enriched biological themes, particularly GO terms
Discover enriched functional-related gene groups
Cluster redundant annotation terms
Visualize genes on BioCarta & KEGG pathway maps
Display related many-genes-to-many-terms on 2-D view.
Search for other functionally related genes not in the list
List interacting proteins
Explore gene names in batch
Link gene-disease associations
Highlight protein functional domains and motifs
Redirect to related literatures
Convert gene identifiers from one type to another.
And more
Sunday, April 3, 2011
Privacy Best Practices - Ethics
- PBP 2005 http://cihr-irsc.gc.ca/e/22085.html
- PBPs Initial Implementation Phase 2005-2007 http://cihr-irsc.gc.ca/e/29138.html
Saturday, April 2, 2011
Identification of cis-regulatory variants that may be causal for disorders
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1618400/
In Silico Detection of Sequence Variations
Modifying Transcriptional Regulation
* https://bora.uib.no/bitstream/1956/2703/1/In_Silico_PLOMed.pdf
* RAVEN (regulatory analysis of variation in enhancers).
* http://www.ploscompbiol.org/article/info:doi%2F10.1371%2Fjournal.pcbi.0040005
* http://www.cisreg.ca
* In this paper we present a Web-based tool for the identification of genetic variation in
potential transcription factor binding sites.
https://depace.med.harvard.edu/work.html
ORegAnno
JASPAR
dbSNP
Phylofoot - Tools for phylogenetic footprinting
sTRAP - http://strap.molgen.mpg.de/
TRANSFAC - transcription factor motifs
A traditional bioinformatics approach to
predict TFBSs is through the application of binding site
profile models known as position-specific weight matrices
(PWMs) [24]. Such matrix models assign a score to each
candidate binding sequence.
Thomas Manke, Matthias Heinig, and Martin Vingron, “Quantifying the effect of sequence variation on regulatory interactions,” Human Mutation 31, no. 4 (April 2010): 477-483.
* sTRAP - http://strap.molgen.mpg.de/
is-rSNP: a novel technique for in silico regulatory SNP detection Macintyre G, Bailey J, Haviv I, Kowalczyk ABioinformatics 2010 26(18): i524. doi:10.1093/bioinformatics/btq378
Predicting functional regulatory polymorphisms [Get the full text PDF from Pubget] Torkamani A, Schork NJBioinformatics 2008 24(16): 1787. doi:10.1093/bioinformatics/btn311
Impact of DNA-binding position variants on yeast gene expression [Get the full text PDF from Pubget] Swamy KBS, Cho C, Chiang S, Tsai ZT, Tsai HNucleic Acids Research 2009 37(21): 6991. doi:10.1093/nar/gkp743
Identification of candidate regulatory SNPs by combination of transcription-factor-binding site prediction, SNP genotyping and haploChIP Ameur A, Rada-Iglesias A, Komorowski J, Wadelius CNucleic Acids Research 2009 37(12): e85. doi:10.1093/nar/gkp381
In Silico Detection of Sequence Variations
Modifying Transcriptional Regulation
* https://bora.uib.no/bitstream/1956/2703/1/In_Silico_PLOMed.pdf
* RAVEN (regulatory analysis of variation in enhancers).
* http://www.ploscompbiol.org/article/info:doi%2F10.1371%2Fjournal.pcbi.0040005
* http://www.cisreg.ca
* In this paper we present a Web-based tool for the identification of genetic variation in
potential transcription factor binding sites.
https://depace.med.harvard.edu/work.html
ORegAnno
JASPAR
dbSNP
Phylofoot - Tools for phylogenetic footprinting
sTRAP - http://strap.molgen.mpg.de/
TRANSFAC - transcription factor motifs
A traditional bioinformatics approach to
predict TFBSs is through the application of binding site
profile models known as position-specific weight matrices
(PWMs) [24]. Such matrix models assign a score to each
candidate binding sequence.
Thomas Manke, Matthias Heinig, and Martin Vingron, “Quantifying the effect of sequence variation on regulatory interactions,” Human Mutation 31, no. 4 (April 2010): 477-483.
* sTRAP - http://strap.molgen.mpg.de/
is-rSNP: a novel technique for in silico regulatory SNP detection Macintyre G, Bailey J, Haviv I, Kowalczyk ABioinformatics 2010 26(18): i524. doi:10.1093/bioinformatics/btq378
Predicting functional regulatory polymorphisms [Get the full text PDF from Pubget] Torkamani A, Schork NJBioinformatics 2008 24(16): 1787. doi:10.1093/bioinformatics/btn311
Impact of DNA-binding position variants on yeast gene expression [Get the full text PDF from Pubget] Swamy KBS, Cho C, Chiang S, Tsai ZT, Tsai HNucleic Acids Research 2009 37(21): 6991. doi:10.1093/nar/gkp743
Identification of candidate regulatory SNPs by combination of transcription-factor-binding site prediction, SNP genotyping and haploChIP Ameur A, Rada-Iglesias A, Komorowski J, Wadelius CNucleic Acids Research 2009 37(12): e85. doi:10.1093/nar/gkp381
Google Summer of Code 2011 Projects
Friday, April 1, 2011
RECOMB 2011
Design of Protein-Protein Interactions with a Novel Ensemble-Based
Scoring Algorithm.
Kyle E. Roberts, Patrick R. Cushing, Prisca Boisguerin, Dean R. Madden
and Bruce R. Donald.
* K*, Protein design, Bruce Donald, Flexible rotamer backbone
* http://ftp.cs.duke.edu/~kroberts/latexProjects/Recomb2011/Final/recomb_calwriteup1.pdf
* http://www.cs.duke.edu/donaldlab/osprey.php
* NSRP - They are synthesized in many bacteria and fungi by large multifunctional proteins called nonribosomal peptide synthetases (NRPS). A unique feature of NRPS system is the ability to synthesize peptides containing proteinogenic as well as non-proteinogenic amino acids. http://linux1.nii.res.in/~zeeshan/nrps.html
* DEE is a provable algorithm, does not produce gaps
* Game theory (minimax) - positive and negative designs
Experiment Specific Expression Patterns.
Tobias Petri, Robert Küffner and Ralf Zimmer.
* Look for genes that deviates from the model 'unexpected genes'
* http://compbio.cs.sfu.ca/recomb2011/recomb2011_submission_249.pdf
* http://www.springerlink.com/content/h725542467jv537j/
Scoring Algorithm.
Kyle E. Roberts, Patrick R. Cushing, Prisca Boisguerin, Dean R. Madden
and Bruce R. Donald.
* K*, Protein design, Bruce Donald, Flexible rotamer backbone
* http://ftp.cs.duke.edu/~kroberts/latexProjects/Recomb2011/Final/recomb_calwriteup1.pdf
* http://www.cs.duke.edu/donaldlab/osprey.php
* NSRP - They are synthesized in many bacteria and fungi by large multifunctional proteins called nonribosomal peptide synthetases (NRPS). A unique feature of NRPS system is the ability to synthesize peptides containing proteinogenic as well as non-proteinogenic amino acids. http://linux1.nii.res.in/~zeeshan/nrps.html
* DEE is a provable algorithm, does not produce gaps
* Game theory (minimax) - positive and negative designs
Experiment Specific Expression Patterns.
Tobias Petri, Robert Küffner and Ralf Zimmer.
* Look for genes that deviates from the model 'unexpected genes'
* http://compbio.cs.sfu.ca/recomb2011/recomb2011_submission_249.pdf
* http://www.springerlink.com/content/h725542467jv537j/
Subscribe to:
Posts (Atom)