microarray data analysis methods

Data normalization

In order to make meaningful comparisons of a gene’s signal intensities between multiple microarrays it is necessary for the signal intensities from each chip to be expressed on the same scale. This is accomplished through data normalization.

The linear normalization provided by the Affymetrix software package MAS 5.0 serves to equilibrate the overall intensity levels of a group of chips, but it does not normalize intensity-dependent differences between chips.  Such differences are only dramatic on a small fraction or arrays and might result from the arrays being scanned such that some of the intensity values are outside the linear range of the detector.  Whatever the cause, we have found that a normalization scheme that can correct for intensity-dependent differences between chips results in a more accurate measure of signal intensity than that accomplished by linear normalization.  Consequently, we have employed a quantile method to normalize Affymetrix GeneChip microarrays.

The quantile normalization method adjusts the signal intensities on each chip as follows.  Within each array the signal intensity of each gene is ranked.  (example) If a set of genes is tied then each is given the mean of the set of ranks that the tied genes span.(example)  Across the arrays, the mean signal intensity for the genes of each rank is calculated.  For every gene on each array the signal intensity of the gene is replaced by the mean signal intensity for the genes of that rank across all of the arrays.

The effect of this normalization method is to make the distribution of signal intensities on each array identical.  The differences in the expression level of a particular gene between arrays are a result of where the measurement is ranked within each array.

Data filtering

The Affymetrix U133A and U133B arrays are capable of detecting the expression of a large fraction of the genes in the human genome.  As we expect that not every gene in the genome is expressed in endothelial cells, we sought to remove from our dataset both those genes that are not expressed in our samples as well as those that might be expressed but for which expression levels could not be reliably quantitated by the Affymetrix system.

To accomplish this task we took advantage of the fact that the probe set for each gene on the Affymetrix arrays contains eleven perfect match (PM) probes and an equal number of mismatch (MM) probes.  Unlike hybridization to the perfect match probes, hybridization to the mismatch probes is non-gene specific and the ratio of PM to MM hybridization indicates whether the PM hybridization is likely to be the result of gene-specific hybridization or is rather likely to be the result of experimental noise.  This concept has been mathematically formalized and algorithms for estimating the probability of PM hybridization resulting from gene-specific hybridization are provided in Affymetrix’s Microarray Suite 5.0 (MAS) software package.  Through the use of adjustable cut offs, MAS can further reduce these probabilities to a trinary “Present”, “Absent” or “Marginal” call – indicating high, low, or intermediate probability of gene-specific hybridization respectively.

We used the Affymetrix-recommended settings in MAS to call whether gene-specific hybridization was detected for each probe set on each array.  We then eliminated from our data set those probe sets for which gene-specific hybridization was “Absent” on every array.  These eliminated genes include both those that are not expressed in our endothelial cell line and those that cannot be detected due to technical limitations.  In our data set we included hybridization-intensity data from probe sets that had at least one “Present” call as we hypothesized that our different treatment conditions might cause expression of some genes to be induced from undetectable levels.

This filtering scheme eliminated xxxx genes from our data set (Table 1.)

Mass Spectrometry 793, Final Paper – Spring 2003

Garrett Frampton

Mass Spectrometry 793

Final Paper – Spring 2003

            It is usually the case that tissue samples subjected to proteomic or genomic analysis are comprised of many different cell types.  Frequently, the cells of interest are interspersed with other cells that the investigator does not want to examine.  These additional cells will certainly reduce the signal of molecules detected from the cells of interest and will likely confound analysis that are focused on only some of the cells in the tissue sample.  The ability to physically select cells of interest out of a tissue section is of great potential utility in many molecular biology applications and mass spectrometry is no exception.  Using laser capture microdissection this is possible.

In laser capture microdissection (LCM) cells can be selected individually out of a thin slice of frozen tissue that is less than one cell (5-20 mm) thick.  In LCM a thin ethylene vinyl acetate thermoplastic film is placed over the prepared tissue sample.  Using a microscope, the investigator targets the cells of interest with a laser.  The laser transiently heats and melts the film, adhering it to the cells below.  This can be repeated to capture as few a one and as many as a few thousand cells.  When the film is removed the captured cells remain adhered and are pulled out of the tissue section.  When performing LCM it is necessary to prepare the sample by actively removing all water in the tissue by treating it with organic solvents such as dehydrated ethanol or xylene.  Nevertheless, laser capture is a relatively gentle procedure, allowing recovery of many proteins and protein complexes, RNA transcripts several kilobases in length, and genomic DNA.

            The principle limitation associated with using laser capture microdissected cells for proteomic and genomic analysis is the small sample size.  Nature has provided a mechanism that allows amplification of single nucleic acid molecules, facilitating genomic analysis of microdissected samples.  Proteins, on the other hand, cannot be amplified.  Fortunately, mass spectrometry is an extremely sensitive detection method, and MS analysis LCM samples is currently viable.  Still, the small quantities of protein analyte that are obtained via LCM are a significant limit to proteomic analysis.

            In has usually been the case that proteins from tissue samples subjected to mass spectrometry are extracted into a lysis buffer.  The solubilized protein is then used for ESI or MALDI MS.  In addition to putting the protein in a state amenable to mass spectrometry, extraction of protein from tissue provides other benefits.  The dissolved protein can be subjected to purification and separation methods that will be discussed later.  Despite the benefits of extracting proteins from tissues, it does result in significant dilution.  Given the small amounts of protein that are procured from LCM samples, the dilution of that protein caused by extraction or separation could prevent detection of a protein of interest.

            Xu et al. describe a method by which cells obtained via LCM can be directly subjected to MALDI MS [1], obviating the need to extract proteins from the sample cells prior to mass spectrometry.  After capturing cells via LCM they attach the thermoplastic film to the target analysis plate and apply sinapinic acid matrix solution directly to the sample.  A portion of the cells’ protein associates with the matrix, making it amenable to ionization.  The sample is then directly subjected to MALDI-TOF mass spectrometry.

            This procedure has advantages to solubilizing protein prior to MS.  The first advantage is that the proteins are present at greater concentrations.  This translates directly to a higher signal to noise in mass spectra, which results in increased sensitivity, increased accuracy, and increased ability to detect proteins of interest.  Another related advantage is that with direct MALDI a very small original sample can be used for MS.  A third advantage is that direct MALDI MS of tissue samples is easier than extracting proteins prior to MS.  Eliminating the extraction creates a protocol that is less time consuming, less costly, less technically challenging, and less variable. 

            In fact, Xu et al. are not the first group to perform direct MALDI of LCM tissue.  Palmer-Troy et al. make an earlier report using a similar technique that they also use to investigate mammary carcinoma [2].  There is a difference between the two techniques with regard to the application of matrix solution to the MALDI target that Xu et al. claim gives their protocol an important advantage.  Palmer-Troy et al. apply 1 ml of matrix solution to the LCM cells on the target plate.  This volume far exceeds the volume of the cells.  In order to restrict matrix volume, Xu et al. use a finely pulled glass capillary to deposit matrix solution on LCM cells under microscopic visualization.  This allows them to employ matrix volumes of 100 pl to 10 nl.  Xu et al. claim that this gives their protocol an advantage because excessive matrix volume can dilute and mix the proteins associated with each cell cluster.  They also state that microspotting matrix solution provides an accurate target for the MALDI instrument’s camera system.  Xu et al. do not show a direct comparison of the two matrix deposition methods, but the spectra in their paper appear to be of higher quality than those of Palmer-Troy et al..  This could be attributed to the fact that Xu et al. are using a higher performance MALDI-TOF instrument though.

            In order to assess the quality of spectra obtained via their direct LCM MALDI MS protocol, Xu et al. perform comparisons between different methods of preparing the same tissue sample.  Their first figure shows that histological staining prior to LCM reduces spectral quality.  They show mass spectra of stained and unstained LCM captured cells.  The unstained cells show much better spectral quality.  This indicates that, if possible, LCM microscopy should be performed without staining.

The second figure states that the tissue dehydration process required in preparation for LCM does not result in significant protein loss from the tissues.  They show spectra obtained from three samples that were unwashed, washed in 70% ethanol for one minute, and washed in water for three minutes prior to direct MALDI MS without LCM.  The spectra appear similar, but this is not an adequate control.  The preparation of tissue for LCM involves not only a 70% ethanol wash, but also (2 x 30 sec) 95% ethanol washes, (2 x 30 sec) 100% ethanol washes, and a (1 x 5 min) xylene wash.  It is necessary to fully dehydrate the tissue in order to get good LCM results.  It seems likely that the stringent dehydration washes, which they do not perform in this control, could cause protein to be lost from the tissue.  Their statement, that the tissue dehydration process required in preparation for LCM does not result in significant protein loss from the tissue, is not tolerably supported.

            The third figure shows the mass spectrum obtained from a LCM sample of ten cells (mouse colon crypt) prepared using their protocol.  The spectrum looks good.  This indicates that their protocol will perform satisfactorily on as few as ten cells.

The fourth figure shows LCM capture of unstained human breast carcinoma.  The images show that almost all of the cancerous cells from the tissue section were removed without affecting the surrounding tissue.  This indicates that staining tissue samples is not required for accurate LCM.  This is important given their prior finding that staining reduces mass spectral quality.

            The fifth figure shows the mass spectra obtained by their direct MALDI MS protocol from microdissected invasive mammary carcinoma and normal breast epithelial cells.  This is a relevant comparison from a biological standpoint because epithelial cells are the primary cells from which breast cancer is derived.  They show spectra over a broad m/z range from 3,000 to 70,000 units.  The spectra appear generally similar with a few dozen clearly defined peaks.  Though the spectra are similar, there are clear differences between them.  There are a few peaks in one of the spectra that appear to be completely absent in the other.  Furthermore, many peaks common to both spectra are observed at significantly different intensities.  This indicates that there are differences between the protein content of these mammary carcinoma and normal epithelial samples, which is not surprising.

Comparison between different types, groups, or treatments of tissue is the basis of proteomic analysis via mass spectrometry.  These analyses generally fall into two categories.  The first type of analysis treats differences between the spectra being compared as markers of the samples being examined.  This can facilitate categorization of unknown samples.  In this case the identity and biological function of those proteins is not important.  The second type of comparative proteomic analysis is concerned with the identity and biology of the changed proteins.  Here the differences between the mass spectra of the samples, along with prior biological knowledge, are used as clues to help learn more about the underlying biology of the samples being examined.  Xu et al. do not comment on any analysis of the differences between the cancer and normal samples.  In this case the comparison between carcinoma and normal tissue provides evidence that their protocols can be used to perform comparative proteomics.

            The protocol described in Xu et al. can be used for rapid whole cell proteomics, but mass spectral analysis of whole cell protein extracts has an important fundamental limitation.  The mass spectra obtained from whole cell protein mixtures, even when acquired on state of the art mass spectrometers, is generally too noisy to detect most proteins of interest.  The most abundant cellular proteins, and their degradation products, obscure detection of other proteins, which are present in quantities many orders of magnitude lower than the abundant proteins.  Consequently, whole cell protein extracts are generally subjected to one or more separation or purification steps in combination with mass spectrometry.  The reduction in complexity by separation or purification can take many different forms.

            Tandem mass spectrometry is the most elegant method of separating proteins for mass spectrometry.  Quadrupole magnets can be used to trap a certain bandwidth of a mass spectrum, filtering a protein mixture by m/z.  In combination with CID, MSⁿ is an excellent method to reduce mixture complexity for whole cell proteomics.

            Chromatography and electrophoresis are two other methods that are commonly used to separate protein mixtures.  They provide excellent results, but cause dilution of samples.  Chromatography and electrophoresis can also be used to separate protein mixtures across multiple dimensions, resulting in further reduction in mixture complexity.

            Affinity purification is another method that can be used to reduce the complexity of whole cell lysates.  In affinity purification, a favorable chemical interaction between an immobilized substrate and the protein(s) of interest is used to pull target molecules out of a complex mixture.  The form and selectivity of affinity capture methods is extremely variable.  Antibodies can be used to capture a particular protein species or a group of proteins.  Phosphorylated proteins can be purified using a variety of methods that convert phosphate groups to affinity handles.  Nucleic acids with sequence dependant specificity can be use to capture proteins.  Numerous inorganic compounds have also been demonstrated to have selective protein affinities.  Affinity purification is the basis of the SELDI technique that will be discussed later.

            Of these separation methods, only tandem mass spectrometry could be integrated into the direct MALDI protocol of Xu et al..  While tandem MS separates gas phase ions within the spectrometer, each of the other methods separates proteins in solution.  Since the protocol of Xu et al. skips the generation of lysate, none of the solution based separation methods would be possible.

            Craven et al. describe a protocol in which proteins are extracted into a lysis buffer and separated prior to mass spectrometry [3].  The protocol of Craven et al. is directly comparable to that of Xu et al. because both groups performed whole cell proteomic analysis of LCM samples, but Craven et al. used 2D-PAGE to simplify sample mixtures prior to MALDI.  They captured several thousand cells from normal kidney, renal carcinoma, and cervix epithelium and extracted proteins into a urea/thiourea-based lysis buffer.  They performed SDS poly-acrylamide gel electrophoresis and used silver staining to detect proteins within the gel.  Each spot was identified and excised by hand, protein subjected to in-gel tryptic digest, and fragment peptides extracted.  This resulted in a dilution of the whole cell protein extract into milliliters of solvent.  The extract from each gel spot was analyzed via MALDI-TOF mass spectrometry and peptide fragment masses were screened against the NCBI database.  Craven et al. identify between 470 and 930 gel features in their various samples, though they do not comment on how many proteins were actually identified.  Additionally, Craven et al. compare the results of using several different stains for LCM imaging, but they do not sample unstained tissue.  Given the evidence presented by Xu et al. that staining negatively affects spectral quality  it would have been good for Craven et al. to have included unstained tissue in this control.

            The 2D-PAGE protocol of Craven et al. generates far more spectral information than the direct MALDI protocol of Xu et al..  Craven et al. detect hundreds of proteins from a whole cell lysate and can obtain enough structural information to identify many of them.  This compares favorably to the protocol of Xu et al., which detects only a few dozen proteins and provides very limited structural information.

            While the 2D-PAGE method clearly has merits, the direct MALDI method also has strengths.  The most appealing feature of the direct MALDI method is it simplicity.  Assuming adequate ionization is achieved, the entire proteome could be ionized directly from tissue.  Compared to the involved chemistry of 2D-SDS-PAGE and MS of tryptic digests, direct MALDI is more likely to preserve proteins in their in vivo state.  Another extremely appealing characteristic of the direct MALDI method is that it allows more reasonable comparisons between proteins signals both within and across samples.  In the 2D-PAGE method each protein is subject to a separate excision, digest, extraction, and MS.  This makes it very difficult to control variability from these steps.  In the direct MALDI method the complexity of the mass spectra serves as an internal control.  Even if the direct MALDI method was naturally extended with MSⁿ in order to detect more proteins, it would still be easier to reconstruct coherent proteome level information than it would be with 2D-PAGE and MS.  It is also important to realize that Craven et al. need to obtain a few thousand cells via LCM before they have enough protein to perform 2D-PAGE.  Xu et al., on the other hand, obtain quality mass spectra with as few as ten cells.  There is no application of LCM that would not benefit from having to obtain 10 cells instead of 10,000.  The tryptic peptide information provided by the 2D-PAGE method is currently invaluable for protein identification, but as knowledge of the proteome grows, identification of proteins without rich structural information will become easier.  As bioinformatics improves, 2D-PAGE and tryptic digest will no longer be necessary to identify most proteins.

            Surface enhanced laser desorption/ionization (SELDI) is another technique that allows reduction of mixture complexity for whole proteome analysis.  It requires relatively small quantities of protein compared with other separation and purification techniques, which makes it attractive for analysis of LCM samples.  In SELDI a complex protein mixture is applied to an affinity surface that subsequently serves as a target for MALDI.  It enables selective protein retention on a wide variety of surfaces.  Chromatographic surfaces include reversed phase, cationic, anionic and metal affinity while bioaffinity surfaces include antibody-antigen, DNA-protein, and receptor-ligand.  Following binding of target proteins to the affinity surface, an appropriate buffer is used to wash away non-specific binding and matrix is applied.  The bound proteins are then subjected to MALDI mass spectrometry.  SELDI is amenable to high throughput applications and can be configured in an array format to run many assays in parallel.  This makes it extremely attractive for many applications.

Batorfi et al. describe a protocol in which SELDI is used for whole cell proteome analysis from LCM tissue [4].  They capture a few thousand trophoblast cells from normal human placenta and hydatidiform mole (a malformed placental feature that generally results in termination of pregnancy) and extract whole protein into a modest 10 ml of lysis buffer.  3 ml of extract is applied to an immobilized metal affinity capture SELDI surfaces.  Following washing, sinapinic acid matrix is applied to the surface and MALDI-TOF MS is performed.  The IMAC surface binds proteins with exposed histidine, tryptophan, or cysteine residues.  This purifies a coherent fraction of the whole cell proteome and puts the bound proteins in a state amenable to mass spectrometry.  While the previous papers did not discuss the analysis of differentially detected ions, Batorfi et al. do find four peaks in their mass spectra that are significantly different between their two experimental groups.  They do not comment on the biological identity or relevance of these proteins.

Of these four papers, only one makes an attempt at the analysis of the biological significance of the data that they have collected.  This is an extremely important issue in current molecular biology.  Many new technologies allow collection of enormous amounts of data, which must be analyzed and deciphered before they are useful.  Unless technological boundaries are being pushed, the collection of proteomic and genomic data without equal investment into the analysis of that data is imprudent.  Considering the problematic nature of cross-experiment comparisons, it is incumbent upon researchers to conduct experiments with data analysis in mind.

Research Proposal

            Each of the methods discussed for conducting proteomic analysis of LCM tissue takes a different approach.  To me, the combination of tandem MS with direct MALDI is the most exciting because of its simplicity.  If I were to propose a research project for a graduate student to extend these findings, I would certainly investigate the application of LCM-MSⁿ.

            LCM of cancerous tissue is particularly appropriate for two reasons.  First, cancerous tissue is a heterogeneous mix of diseased and normal tissue.  Analyzing the cancer cells separately might be necessary to figure out the molecular pathology of cancer.   Second, cancerous tissue is frequently available and diagnostic testing of cancer biopsies would both practical and valuable.  Consequently cancer tissue is a great starting point to investigate LCM-MSⁿ.  Mammary carcinoma would be a good system because samples are available and the cell type of origin is known.

            Pushing the limits of LCM-MSⁿ would require repeated analysis of the same samples.  To that end I would capture via LCM many dozen samples each of 100 cells from carcinoma and normal epithelial cells and freeze the samples at -80ºC for future use.  This could be problematic because long-term storage of LCM cells may result in protein degradation.  It would be important to monitor this factor and fresh LCM samples might have to be used.  I would also make sure that the original tissue was fresh and handled well in order to alleviate to problem of protein degradation prior to LCM.

            In order to perform tandem MS, it is necessary to define the desired bandwidth of m/z that will be captured in the first sector of the instrument.  There are two approaches to this problem.  One is to scan the entire range of m/z and subject each interval to a second stage of MS.  The second method is to use prior knowledge to select only those bands of m/z that contain important information.  I would investigate the scanning approach because it would be applicable to any type of sample.

The main problem with the scanning approach to tandem MS is that it requires collection of a separate CID mass spectrum at every bandwidth scanned.  This would result in thousands of spectra and hundreds of thousands of MALDI shots for a single 2D mass spectrum.  This would be difficult in terms of both sample size and time.  In order to efficiently investigate high-resolution tandem MS I would limit my investigation to a small portion of the initial mass spectrum.  I would choose a range of a few hundred m/z units, which contained a complex spectrum and possibly contained targets of interest that have been identified by other investigators.

The goal of this work would be to obtain a good 2D mass spectrum, over a limited range of initial m/z, using direct MALDI of LCM tissue.  The separation provided by tandem MS with CID would allow resolution of far more molecular species than in the report of Xu et al..  This would be an important extension of the direct MALDI ionization protocol because separation of complex protein extracts is critical for whole cell proteomics.
References

1.      Xu B, Caprioli R, Sanders M, Jensen R: Direct Analysis of Laser Capture Microdissected Cells by MALDI Mass Spectrometry. American Society of Mass Spectrometry 2002, 13:1292-1297

2.      Palmer-Troy DE, Sarracino D, Sgroi D, LeVangie R, Leopold P: Direct Acquisition of Matrix-assisted Laser Desorption/Ionization Time-of-Flight Mass Spectra from Laser Capture Microdissected Tissues. Clinical Chemistry 2000, 46:1513-1516

3.      Craven R, Totty N, Harnden P, Selby P, Banks R: Laser Capture Microdissection and Two-Dimensional Polyacrylamide Gel Electrophoresis.  American Journal of Pathology 2002, 160:815-822

4.      C Batorfi J, Ye B, Mok S, Cseh I, Berkowitz R, Fulop V: Protein profiling of complete mole and normal placenta using ProteinChip analysis on laser capture microdissected cells. Gynecologic Oncology 2003, 88:424-428

more random text

Embryonic stem cells or ES cells are cell lines derived from the inner cell mast of blastocyst stage embryos.  ES cells have two key properties that make them an ideal model system to study human development.  The first, self-renewal, is the ability to replicate almost indefinitely in cell culture.  This allows researchers to obtain large and homogenous populations of cells to study.  The second property, pluripotency, is the ability to differentiate into all of the of different cell types that make up an adult organism.

The science of regenerative medicine hopes one day to discover how to create healthy tissues to replace those that lose function due to disease, injury, or age. Since ES cells have the developmental potential to differentiate into any tissue are believed to hold great promise for regenerative medicine

In 2006 groundbreaking experiments by Takahashi and Yamanaka demonstrated that pluripotent cells could be created from somatic cells by forced expression of the ES cell transcription factors Oct4, Sox2, Klf4 and cMyc. These induced pluripotent stem cells, or iPS cells, are highly similar to embryonic stem cells.

. The generation of iPS cells was a huge breakthrough in regenerative medicine because they are similar to embryonic stem cells and can be derived in a patient-specific manner, from adult somatic cells, requiring no embryonic tissue. 

Since the initial reprogramming experiments in 2006, the pace of research in this field has been very rapid. In chimera experiments iPS were show to be able to give rise to germ line cells. Then in tetraploid complementation assays, it was demonstrated iPS cells can contribute to all of the tissues that make up an adult organism Since their initial generation in mouse, iPS cells have been made in several species including human .  They have been made from a large number of different starting cell types and using many different sets of ES cell transcription factors.

One of the original reprogramming transcription factors, Myc, is a known oncogene.  It has since been shown that iPS cells can be made without the use of myc. It has also been shown that iPS cells can be made with viral vectors than can later be excised from the genome or with vectors that do not integrate into the genome at all. iPS cells have also been made using small molecules and proteins as reprogramming agents.

Finally, a number of in vitro disease models have been made using iPS cells. For example, iPS cells were made from patients suffering from Parkinson's disease and then differentiated in vitro into neuronal cells. Since these neurononal cells have the genome of a person who developed Parkinson’s disease, they may be an important tool for understanding the disease pathology.

As I said earlier ES cells and iPS cells are highly similar. They express the same markers of pluripotency and have the same cell morphology. They also behave similarly in a broad range of phenotypic assays including            - embroid body formation           - teratoma formation - germline transmission             - and tetraploid complementation   Whether ES cells and iPS cell are truly equivalent is a very important question, since if they are equivalent, then the knowledge that has been gained in years of embryonic stem cell research will be applicable to iPS cells as well.

Recently, several studies have been published saying that there are differences in the gene expression programs of ES and iPS cells. If it is true, that ES and iPS cells have different gene expression programs, then this indicates that they are not equivalent cell types. This has huge implications for the entire field of stem cell research. If ES and iPS cells are not equivalent, then iPS cells might not hold the promise for regenerative medicine that we had hoped.

We decided that it would be critical to answer for ourselves whether ES and iPS cells are equivalent cell types.  We reasoned that obtaining genome-wide maps of H3K4me3 and H3K27me3 histone modifications in addition to gene expression profiles for a panel of ES and iPS cell lines would allow us to obtain a highly detailed and quantitative assessment of both the current transcriptional state of ES and iPS cells as well as their future developmental potential.

WE decided chose to profile the H3K4me3 and H3K27me3 histone modications because that are two of the most well studied histone modifications.  H3K4me3 is generally associated active genes while H3K27me3 is associated with genes that are not being transcribed.  In ES cells active genes have the K4 mark at their transcription start sites.  There is another class of genes that are occupied by both K4 and K27. These genes are especially interesting because they are include many of the most important genes for controlling differentiated cell types.

Homeobox transcription factors for example are nearly all occupied by both K4 and K27 in ES cells. These genes with both K4 and K27 are not expressed in ES cells, otherwise they would promote differentiation, but they are poised for later expression. In a differentiated cell that expresses one of these genes, the K27 mark is lost while K4 is retained.  In a differentiated cell that turns off one of these genes, the K4 mark will be lost and K27 is retained. At some genes, both K4 and K27 are retained which means that this gene is still poised for expression as the cell differentiates further.Together the genome wide locations of the K4 and K27 marks reflect both of the current transcriptional state of a cell as well as it future developmental potential.

We reasoned …

By comparing these data we can accurately asses whether there are truly differences between ES and iPS cells.  We assembled a collection of 6 independent ES cell lines each from separate donors, and 6 iPS lines, 4 derived from one fibroblast donor and 2 from another donor.   We also examined fibroblast cells as a control. 

Together H3K4me3 and H3K27me3 location analysis and microarray based gene expression experiments allow us to get a highly quantitative measure of cell state  This reflects both the current transcriptional state of the cell as well as its future developmental potential.  

MORE RANDOM GENOMICS TEXT

Methods for Comparison of ChIP-Seq experiments

Since the ChIP-Seq technique was first developed in 2007, a lot of work has gone into developing methods for analyzing this data.  Dozens of papers and several reviews have been published describing methods and tools for the analysis of individual ChIP-Seq experiments (references), but there has been far less effort devoted to methods for comparing the results from multiple ChIP-Seq location analysis experiments.  A comparison between multiple ChIP-Seq experiments might be necessary for examining the occupancy of a single factor in multiple conditions or for comparing the occupancy of different factors in the same cell.

Two of the methods that are commonly used to compared the results of multiple ChIP-Sqee experiments are Venn Diagrams and clsutergrams (Figure ?). Venn diagrams show the number of enriched regions or genes that overlap between a set of ChIP-Seq datasets. One weakness of Venn diagram analysis is that it requires determining a threshold for enrichment in each dataset a priori, which is often problematic. Additionally, the Venn diagram analysis frequently underrepresentats the similarity between two different experiments.  Two experiments that appear to be very similar might only be 50% overlapping in a Venn diagram analysis, and it is rare to observe two datasets overlapping more than 80%.  Lsutergrams show this and that etv.  The primary weakness of clsutergrams is tthey tend to highlight simialrities between datasets. 

Magnitude of enrichement more important than hypergeometricc significance

Framptongram - cytoscape

                        Methods for es v sips compairosn

                        Normalization and Comparitive Analysis

No threshold, preserve order

http://functional-strength.blogspot.com/2010/11/lessons-learned-as-adolescent-and-adult.html

http://functional-strength.blogspot.com/2010/11/lessons-learned-as-adolescent-and-adult.html

funness

things are fun to do every once it a while, but you suck at them

things you do alot, you are good at, but there are no fun.

what a pain.

One time, for real though, I did put 2 Q-tips in my ear at the same time and I took a shit

"One time, for real though, I did put 2 Q-tips in my ear at the same time and I took a shit.  It was the best feeling ever."

-Joe Rogan

Green Jolly Ranchers

Green Jolly Ranchers taste like soap.  I wish they would make a new flavor.

ideas

It is not possible to live in cleanliness 
Having property taxes fund public schools is wrong 
It is not reasonable to expect the best healthcare that money can buy, unless you are willing to pay for it. 
Don't borrow money to buy things that you do not need. 
Long term prison sentences don't make any sense. A sentence of up to a few years is a practical punishment. A life sentence for someone who cannot remain in society makes sense. What is a 30 year sentence supposed to do? 
We should spend as much resources as neccesary to assure that all children are as well educated as is practical. 

Biology

The function of a diploid genome is poorly understood. Intereactions between spefici allele combinations must be prominant, but this generally unappreciated.

Energy

Reusing and repairing things is fundamentally better than making new ones, even if it requires more immediate resource to rebuild or repair.
At 1 kw / square meter solar energy, 10% efficency, 8 hours per day of sunlight, each year a 1 square meter photovoltaic panel will produce the same amount of electric energy as the heat energy content of seven gallons of gasoline.
1 gallon of gasoline contains 41 kw hours of heat energy
1 horsepower is 746 watts
Wireless transmission of eletric energy would be a huge invention
Hydrogen is not an energy source
There are 100 million men in the USA that flush when they urinate. If they pissed in the sink and ran 1 pint of water down the sink instead of using 1 gallon of water in a flush toilet, 1 urinations a day, 365 days a year, it would save 32 billion gallons of water and sewage.
Nuclear power is not sustainable because the problem of radioactive waste has not been solved.

ES cells

Human embryonic stem cells and reprogrammed cells virtually identical CAMBRIDGE, Mass. (August 5, 2010) – Human embryonic stem (ES) cells and adult cells reprogrammed to an embryonic stem cell-like state—so-called induced pluripotent stem or iPS cells—exhibit very few differences in their gene expression signatures and are nearly indistinguishable in their chromatin state, according to Whitehead Institute researchers. Their results are published in the August 6 issue of Cell Stem Cell. iPS cells are made by introducing three key genes into adult cells. These reprogramming factors push the cells from a mature state to a more flexible embryonic stem cell-like state. Like ES cells, iPS cells can then, in theory, be coaxed to mature into almost any type of cell in the body. Unlike ES cells, iPS cells taken from a patient are not likely to be rejected by that patient’s immune system. This difference overcomes a major hurdle in regenerative medicine. "At this stage, we can’t yet prove that they are absolutely identical, but the available technology doesn’t reveal differences," says Whitehead Institute Member Richard Young. "It does mean that iPS cells could be useful as personal ES cells in the future." “Billions of dollars have been invested in the idea that we will use ES cells at some point in the future as therapeutic or regenerative agents, but for ethical and practical issues, this may not be possible,” says Garrett Frampton, a co-first author on the Cell Stem Cell paper and a graduate student in the lab of Whitehead MemberRichard Young. “But if they work out therapies with ES cells, and iPS cells are equivalent to ES cells, then the idea is that those therapies could be used with iPS cells as well. Whereas if iPS cells are different from ES cells, then who knows if you can use iPS cells for therapy?” Since iPS cells were first developed in 2006, the similarities and differences between ES and iPS cells have been hotly debated in the scientific community. Thus far, researchers have gauged the cells’ equivalence by determining whether the cells express the same genes, but such studies have yielded mixed results. In revisiting the question of the cells’ equivalence, Frampton and co-first author Matthew Guenther, who is a scientist in the Young lab, analyzed gene expression patterns and the cells’ chromatin structure. Chromatin is the packaging of DNA around a protein scaffold. Variations in chromatin “packaging” can themselves alter gene expression, yet Guenther and Frampton found that human iPS and ES cells to be almost identical in both gene expression and chromatin structure. “At this stage, we can’t yet prove that they are absolutely identical, but the available technology doesn’t reveal differences,” says Young, who is also a biology professor at MIT. “It does mean that iPS cells could be useful as personal ES cells in the future.” Some earlier studies have indicated that iPS and ES cells are dissimilar enough to be classified as different cell types. To see why the results differed so strikingly from theirs, Guenther and Frampton reanalyzed those studies’ data. They concluded that the differences noted in other studies were not consistent between different laboratories and thus were not likely to be a result of fundamental differences between the cell types. “The key question is, are any of these differences functionally relevant? Do they change how a cell matures or not?” says Whitehead Member Rudolf Jaenisch, whose lab worked closely with Guenther and Frampton. “The earlier documented differences were more noise than anything. But other tests may give you a different answer. So it is still an open question, something that the field will continue to struggle with and have to decide.” Guenther agrees. “Our paper addresses the ground state of iPS and ES cells in a laboratory setting,” he says. “But we don’t know for a fact that they won’t behave differently when they mature into various cell types or tissues. That’s the next step.” This research was supported by donations from Liliana and Hillel Bachrach, Landon Clay, and Susan Whitehead.
	Written by Nicole Giese. * * *  Rudolf Jaenisch's primary affiliation is with Whitehead Institute for Biomedical Research, where his laboratory is located and all his research is conducted. He is also a professor of biology at Massachusetts Institute of Technology. Richard Young’s primary affiliation is with Whitehead Institute for Biomedical Research, where his laboratory is located and all his research is conducted. He is also a professor of biology at Massachusetts Institute of Technology. * * * Full Citation: “Chromatin Structure and Gene Expression Programs of Human Embryonic and Induced Pluripotent Stem Cells” Cell Stem Cell, August 6, 2010 Matthew G. Guenther (1,3), Garrett M. Frampton (1,2,3), Frank Soldner (1), Dirk Hockemeyer (1), Maya Mitalipova (1), Rudolf Jaenisch (1,2), and Richard A. Young (1, 2). 1. Whitehead Institute for Biomedical Research, 9 Cambridge Center, Cambridge, Massachusetts 02142, USA 2. Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA 3. These authors contributed equally. Whitehead Institute for Biomedical Research is a nonprofit, independent research and educational institution. Wholly independent in its governance, finances and research programs, Whitehead shares a close affiliation with Massachusetts Institute of Technology through its faculty, who hold joint MIT appointments.