Mapping Gene Regulatiry Networks in Yeast using DNA Microarrays, proteomics

and genmapp.

Oitvai Barabasi 2002 neat graphic showing different levels of cell organization.

cold shock (because it's illstudied so far)

GenMAPP, draw pathway, color

www.genmapp.org

cold shock (because it's illstudied so far)

I wanted to make sure that I can build Qt apps on os x, the documentation is a little bit weak, especially with respect to explaing how to get qmake to automatically shove the right framework components into the .app bundle.

However the following steps were able to build an appropriate bundle.

export QTDIR=/usr/local/Trolltech/Qt-4.1.0 export APP=hello mkdir ${APP}/Contents/Frameworks rsync -av --exclude "_debug" --exclude "Headers" ${QTDIR}/lib/QtCore.framework ${APP}.app/Contents/Frameworks rsync -av --exclude "_debug" --exclude "Headers" ${QTDIR}/lib/QtGui.framework ${APP}.app/Contents/Frameworks rsync -av --exclude "_debug" --exclude "Headers" ${QTDIR}/lib/QtOpenGL.framework ${APP}.app/Contents/Frameworks

install_name_tool -id @executable_path/../Frameworks/QtCore.framework/Versions/4.0/QtCore ${APP}.app/Contents/Frameworks/QtCore.framework/Versions/4.0/QtCore install_name_tool -id @executable_path/../Frameworks/QtGui.framework/Versions/4.0/QtGui ${APP}.app/Contents/Frameworks/QtGui.framework/Versions/4.0/QtGui install_name_tool -id @executable_path/../Frameworks/QtOpenGL.framework/Versions/4.0/QtOpenGL ${APP}.app/Contents/Frameworks/QtOpenGL.framework/Versions/4.0/QtOpenGL

install_name_tool -change ${QTDIR}/lib/QtCore.framework/Versions/4.0/QtCore @executable_path/../Frameworks/QtCore.framework/Versions/4.0/QtCore ${APP}.app/Contents/MacOS/${APP} install_name_tool -change ${QTDIR}/lib/QtCore.framework/Versions/4.0/QtGui @executable_path/../Frameworks/QtGui.framework/Versions/4.0/QtGui ${APP}.app/Contents/MacOS/${APP} install_name_tool -change ${QTDIR}/lib/QtCore.framework/Versions/4.0/QtOpenGL @executable_path/../Frameworks/QtOpenGL.framework/Versions/4.0/QtOpenGL ${APP}.app/Contents/MacOS/${APP}

install_name_tool -change ${QTDIR}/lib/QtCore.framework/Versions/4.0/QtCore @executable_path/../Frameworks/QtCore.framework/Versions/4.0/QtCore ${APP}.app/Contents/Frameworks/QtGui.framework/Versions/4.0/QtGui install_name_tool -change ${QTDIR}/lib/QtCore.framework/Versions/4.0/QtCore @executable_path/../Frameworks/QtCore.framework/Versions/4.0/QtCore ${APP}.app/Contents/Frameworks/QtOpenGL.framework/Versions/4.0/QtOpenGL

There's probably some way of emedding that in a .pro file, but I don't know it yet. Also the cp -R really should be modified to not copy the header and debug parts of the framework if one wants a reasonable download size.

need multiple screenshots of larger myogenen pathway

need multiple genomes, cistamatic might have some

dog, mouse, human, opossum, chicken, fish, fugo,

old mouse genome, had some issues

worry about holes in the genome, can cause drops of regions?

whats a core genome? is it quality of genome, or is it evolutionary distance?

go down to next gene go up 50 kb/then overlapping modules including the next/

Eric showed up to advertise his stochastic grammer techonology. http://computableplant.ics.uci.edu/papers/StochProcSemanticsTR.pdf

It appears to be a production rule -> result with a probablity Of course the various components can be mutualy dependent on various state parameters.

The question is can this be used for biological network evolution?

E.g. simplified cell cycle?

Also Eric's papers are linked from http://www.ics.uci.edu/~emj/

I've been digging through ~kent/src looking to see how the UCSC browser works, so we can write our own API for grabbing data out of it.

Table descriptions from from .as files (autoSql) and are loaded into tableDescriptions.

There's grp table which lists the defined groups.

The map between tables and which group they belong to appears to come from *.ra files. I haven't found a place in the database where they're stored yet.

System Message: WARNING/2 (<string>, line 10); backlink

Inline emphasis start-string without end-string.

We're trying to determine if there are repeats near genomic probems.

The first pass file had a large number of repeats in it.

high_genomic_signal_fewer_repeats.txt had those stripped out.

I created a fasta file with the following python code

data = [ x.strip().split('\t') for x in open('high_genomic_signal_fewer_repeats.txt','r')]
open('high_genomic_signal_fewer_repeats.fa','w').writelines([ ">%s|%s\n%s\n" % (x[2],x[0],x[3]) for x in data])

ok yeah I shouldn't do that in so few lines.

Hopefully the following

blastall -p blastn -e 0.000001 -d mouse_34.0.fa -i high_genomic_signal_fewer_repeats.fa -m 7 | tee high_genomic_signal_fewer_repeats.blast.xml

command will generate a new, smaller xml results file.

• Subversion

  The full documentation for subversion can be found at http://svnbook.red-bean.com/ However here's a quick introduction.

  Subversion is a version control system designed to fix several issues that were found in cvs, such as being able to rename files and process binary files more efficiently.

  Version control systems allow one to keep track of how a file changes over time, and help multiple people who are modifying the same file. For instance if two people had copies of a file and each independently added some text, if they wanted to merge the file one of them would have to take the others file and attempt to identify the differences and merge the two together.

  Subversion, however, would attempt to merge the two chunks automatically. It does this by seeing if the changes touch the same line of text, if they do it will generate a conflict, otherwise it will go ahead and merge them.

  It will also help you roll back to a previous change, or compare how a file has changed over the past few months.

• Subversion Quick Start

  svn co <svn url> [destination] (checks out a subversion url to either the name of the last part of the url or the optional destination name).

  svn update when run in a project directory will update to the latest version of that project.

  svn status will show you what changes you've made to the directory try

  svn status -u will check the network to see if there are an updates

  svn commit [filename filename ...] will commit either everything (when there are no files, or the specific file list to the archive)

• OS X Subversion

  Command line copy of subversion built for OS X is available at http://metissian.com/projects/macosx/subversion/ or http://www.codingmonkeys.de/mbo/

  Documentation on installing a Mac OS X subversion client is availabe at

  http://weblogs.uncc.edu/blog/jwedgeco/2004/08/17/mac_os_x_subversion_client_works

• asked about biohub
• show list of orthologs
• links to expression array
• links to motifs
• link to structure
• show data source
• map viewer

So I read through quite a bit of the samba howto and finally got samba setup.

One big problem I ran into was the following

operation is not possible without initialized secure memory (you may have used the wrong program for this task) [2005/08/16 14:24:27, 0] smbd/server.c:main(798)

It appears that happens when I have SSL or TLS initialized, I solved it by upgrading libldap2 from 2.1.23-1 to 2.1.30-8 and libgnutls7 from 0.8.12-3.1 to 0.8.12-7

• Getting ldap auth working

apt-get install libnss-ldap pam-ldap

Still have some big problems in that i have too many different password databases floating around.

$ ldapsearch -H ldaps://127.0.0.1/ -b dc=mydomain -x ldap_bind: Can't contact LDAP server (-1)

It turns out that the problem was in my /etc/ldap.conf file

http://www.openldap.org/lists/openldap-software/200409/msg00060.html

suggested using -d 1

The interesting line was

TLS: could not load verify locations (file:/etc/ldap/slapd.pem',dir:').

Ah! I'd renamed that file... so it was trying to validate the self signed server cert which wasn't working... ok

However the openldap howto http://www.openldap.org/pub/ksoper/OpenLDAP_TLS_howto.html recommended putting both the public and private key into the same file. Since it appears i need to distribute the key to my client machines it'd be really nice if they were seperate.

Ok now I get it...

In the ldap.conf file you need to make sure that TLS_CACERT is pointing to either the certificate (for self signed certs) or the certificate authority chain (for trusted certs)

Note it's also important for the base to be set properly in /etc/ldap/ldap.conf (it certainly cuts down on the amount of typing one needs to do)

To get apache to work I needed to add LDAPTrustedCA /etc/ldap/mus-slapd.crt.pem LDAPTrustedCAType BASE64_FILE to get apache to me happy with my cert, and i still think its using ssl and not tls.

installed pam_ldap and nss_ldap, they both wanted a proxy user for authentication, and also the admin user for changing passwords.

The two differnt types of PCAGinsu plots are starting from either 0 or 1.

They should be consistent.

I was originally thinking that the algorithm should look down its list of algorithms and modify its paramemters. Though perhaps instead the wrapper should have some call that allows one to set the next lower algorithms parameters which would then know if it needed to delegate it further down.

So the default color encoding for matplotlib appears to be RGB where the rgb values are floats in the range [0,1], you can pass in larger values, 20,0,0 still makes red, while 255,0,0 makes black. (Which I don't understand.)

The color model being used previously looks to be some version of HSV so I may need some mapping layer to fix this...

Squished a few more bugs, i modified all of the create_plot functions to grab the iplot filename so it doesn't try writing out a new file each time it generates the plot.

However I'm still getting two generations for PCA projection because of the get_scatter_axis. I'm wondering if I should just keep the iplot/matplotlib object in memory. (though that breaks serialization, though I could recreate the plot when it reloads?)

• Working on tracy's oligoselect program
  • algorith apperes to be:
    • starting at the 3' end, test each window to see if it's a valid candidate
    • the candidate test is:
      • is the gc content within the specified range
      • is there any stretch of a single base pair type > 25%
      • does any bp make more than 50% of the sequence
      • is it self annealing?
      • is the blast score < percent match

I'm not generating the client side image map for the datapoints yet.

One thing to remember is trying to decide between passing around a row id for generating the plot view like chris' code vs passing around the a label from the cluster labeling.

Also is it safe to be letting people use random values for doing a labeling lookup?

Though I'm still wondering how we can break free of the JPL code and have something that's fully open source.

Other than that I do need to finish getting it ready for OS X

The dirtiest was chaing the system file and commenting out the #ifndef #endif around #include