Home Papers Reports Projects Code Fragments Dissertations Presentations Posters Proposals Lectures given Course notes

Pattern Analysis of Human Histone Methylations and Acetylations through Cross Correlation Maps

Werner Van Belle1* - werner@yellowcouch.org, werner.van.belle@gmail.com

1- Department for Biosystems Science and Engineering; ETH Zurich; Mattenstrasse 24, Building 1058, Basel; Switzerland

Abstract :  This paper investigates how to properly calculate relations between multiple genomic tracks produced by deep sequencing technology. Based on a cross-correlation analysis we measure relations between peak and peaks, peaks and areas and areas and areas over short or long distances. Using this method we revisit previously published acetylation and methylation patterns and present 41 maps, one for each of the measured histone modifications.

Keywords:  histone methylation, acetylation, cross correlation, peak detection, epigenetics
Reference:  Werner Van Belle; Pattern Analysis of Human Histone Methylations and Acetylations through Cross Correlation Maps; Bioinformatics; YellowCouch; February 2009
See also:
The research article.
The browsable correlation maps between Histone Methylation and Acetylation tracks
Some posters [1, 2, 3]
Presentations of this research at FMI, Novartis, UNIL, University Zürich and ETH Zürich [1, 2]

1 Introduction

Recent advances in sequencing technology made it possible to sequence large amounts of a variety of input material. Among these is the possibility to measure protein DNA binding positions through chromatin immunoprecipitation (CHIP). Zhao, Wang, Barski and others performed such experiments and generated tracks for a large collection of human histone methylation and acetylation modifications [1, 2]. This amounted to around $126.10^{9}$ bases, which poses interesting analysis challenges. One of the modern questions regarding such signal tracks is how various patterns within one track correlate with patterns in another track.

Current strategies to analyze this type of data consists of trying to develop peak-finders to find binding positions and then find some meaningful relation between the occurrence of a peak and the presence of transcription for instance. Although this approach seems simple, it is essentially an object recognition problem and it is not at all clear that a simple model of Gaussian peaks is actually a proper one. Techniques such as deconvolution and hidden Markov models might be appropriate, it is however premature to decide on which underlying signal model is a correct one. Furthermore, converting this type of profile to a collection of peaks makes certain statistical procedures later on quite complicated. It is for instance unclear how peaks can be correlated between different tracks without specifying an arbitrary factor that specifies when two peaks are sufficiently similar (Stated differently: what is the minimum allowed shift before two peaks are treated as distinct peaks ?).

Another strategy often employed to analyze this type of data is to rely on known information and then partitioning the data into subsets or regions and then compare average signals across these subsets. For instance in many papers the genomic track is immediately divided in active and inactive genes after which profiles from these different types of classes are compared. When one tries to plot error bars on this type of plots, one quickly discovers the limitations of this approach since the error bars are completely off the chart.

Further scientific errors are often made:

  1. Many observations and claims are not correctly tested (Correlations for instance)
  2. Often demonstrations are used to prove some correlation (high versus low activity plots)
  3. Often 'representative' samples are used to give a wrong impression on the overall data.
In this paper we were also interested to understand relations between different tracks, but we preferred not to rely on too many assumptions on the type of 'objects' we expected to find. Essentially, we did not expect to find only peaks. Instead we preferred to work with the signal as measured and see how different areas correlated against each other. This is obviously a computational intensive task, but it can offer important insights into the structure of the data and whether certain yet unknown systematic factors are present. We discuss a number of them later on.

2 Experimental Setup

We downloaded the methylation and acetylation tracks with accession numbers SRA00206 [1] and [2] from the short read archive at NCBI [3]. The 20 methylations are $H_{2}BK_{5}me_{1}$, $H_{3}K_{27}me_{1}$, $H_{3}K_{27}me_{2}$, $H_{3}K_{27}me_{3}$, $H_{3}K_{36}me_{1}$, $H_{3}K_{36}me_{3}$, $H_{3}K_{4}me_{1}$, $H_{3}K_{4}me_{2}$, $H_{3}K_{4}me_{3}$, $H_{3}K_{79}me_{1}$, $H_{3}K_{79}me_{2}$, $H_{3}K_{79}me_{3}$, $H_{3}K_{9}me_{1}$, $H_{3}K_{9}me_{2}$, $H_{3}K_{9}me_{3}$, $H_{3}R_{2}me_{1}$, $H_{3}R_{2}me_{2}$, $H_{4}R_{3}me_{2}$, $H_{4}K_{20}me_{1}$ and $H_{4}K_{20}me_{2}$. The 18 acetylations are $H_{2}AK_{5}$, $H_{2}AK_{9}$, $H_{2}BK_{120}$ , $H_{2}BK_{12}$, $H_{2}BK_{20}$, $H_{2}BK_{5}$, $H_{3}K_{14}$, $H_{3}K_{18}$, $H_{3}K_{23}$, $H_{3}K_{27}$, $H_{3}K_{36}$, $H_{3}K_{4}$, $H_{3}K_{9}$, $H_{4}K_{12}$, $H_{4}K_{16}$, $H_{4}K_{5}$, $H_{4}K_{8}$ and $H_{4}K_{91}$. Aside from these there were also tracks for $CTCF$, $Pol-II$ and $H_{2}AZ$. Sadly enough there were no RNA-SEQ tracks for the specific samples.

The original data size of all these tracks is around bases, which takes around 231 GB. We aligned each of these tracks to the Human genome using a non standard invocation of Eland [4] such that it would also produce position information for hits aligning to multiple positions (multi-hits). The alignment files took around 35GB.

From these alignments on, we generated a genome wide signal track for each of the different methylations and acetylations. This was done through a modified version of Eland2Wig ([5], which was able to handle the specialized output of the multi-hit alignment. Initially, each multi-hit was fully counted at each of its aligning positions (weight = 1), although in later situations we experimented with a weighing factor of $1\slash n$ and a weighing factor of 0. The last option is also the standard behavior of the Illumina pipeline. Each alignment start position was then extended, thereby taking the alignment direction into account. We used two different approached. Originally we used only the read length (24, 25, 26 or 35, 36, depending on the experiment), such that each position in the genome would reflect how many times that base was sequenced. A second approach relied on extending each alignment up to the average fragment length (160 base pairs)as put on the flowcell. So the signal tracks would consequently reflect how many times a specific base was present in the sample.

To compare the signals surrounding transcript start sites across methylations/acetylations we extracted little vectors (in the mathematical sense) of around 2000 floating points and stored them separately on disk . The transcription start sites were taken from the Ensembl database (homo_sapiens_core_47_36i) and resulted in around 22000 transcripts. For each transcript we generated three different vectors. Each of these vectors would have the first half (the first 1000 bases) reflect the upstream region and the second half would be one of the following three downstream regions.

  1. The first downstream region was the transcript itself. Hereby we simply removed all of the exons and scaled the full length of the transcript to the 1000 bases required for the second half of the signal vector.
  2. The second downstream extraction consisted of an alternation of exons and introns, up to 10 exons. Each exon or intron was scaled to a length of 52 bases and placed behind each other in the signal vector. Areas without a signal (this occurs for transcripts with less than 10 exons) were annotated as 'unknown'.
  3. The third downstream extraction was a copy of the downstream genomic region, disregarding any exon or intron information.

Figure 1: Signal areas around transcription start sites are extracted respecting the transcription direction after removal of overlapping transcripts. Each extraction includes a 1000 basepair upstream region and a 1000 basepair downstream region. For each area we have three possible downstream regions: A) Once with only the full gene transcript scaled to 1000 bases, B) once with an alteration of exons and introns, each exon/intron scaled to 52 bases and C) once with the proper 1000 bases of the downstream genomic area. For each area we also produced a non-normalized (absolute) variant and a normalized (relative) variant where the total mass of the area would become 1, regardless of the absolute signal strength.

Figure 1 illustrates the different extractions. For each of the extractions, the transcription direction was taken into account and we removed all transcripts that would overlap with different genes.

A last aspect on the TSS Area extraction was the question whether we should normalize each of the vectors. If we would normalize each TSA such that the total sum would be the same, then each transcription area would have the same weight in subsequent analysis steps. Whether normalization or not is appropriate in this type of analysis is still unclear. We experimented with both approaches.

All these different considerations lead to 12 different data sets of transcription areas. One for each combination of area normalization (absolute versus relative), fragment extension (to the read length or to the selected 160 bases) and transcription area extraction (transcript only versus exon/intron alternations versus the genomic signal). These are summarized in table 1.

Table 1: We produced 12 different data collections of transcription areas. One for each combination of area normalization (absolute versus relative), fragment extension (to the read length or to the selected 160 bases) and transcription area extraction (transcript only versus exon/intron alternations versus the genomic signal). The text within each cell reflects the name the data set got; eg. ta.0; tf.160 etcetera.
Area Fragment Transcription Area Extraction
Normalization Extention Transcript Exon/Inron Alternations Genomic
Relative read-length ta.0 tb.0 tc.0
`` 160 ta.160 tb.160 tc.160
Absolute read-length td.0 te.0 tf.0
`` 160 td.160 te.160 tf.160

3 Algorithm

3.1 Correlation

Figure 2: A recurring question with this type of genomic signal tracks is whether the signal in one track correlates with the signal in another track. For instance whether the transcription start site signal of one track correlates with the transcription start site signal in another track. More difficult questions involve translations where one asks: does the signal at TSS-100 in one track correlate with the signal at TSS+200 in another track ?

A typical question when dealing with this type of data is whether specific areas around the transcription start site (TSS) correlate with each other (Figure 2). One can for instance have an RNA track (obtained with RNA-SEQ) and compare it against a DNA protein binding track (CHIP-SEQ), or one could ask the more specific question whether $H_{3}K_{27}ac$ correlates against $Pol-II$ at the transcription start site. Obviously the next question would be whether such a correlation exists before or after the transcription start site. To approach this problem we started to systematically correlate each position around the transcription start site from one track against each potential position around the TSS in the second track.

This process is illustrated in Figure 3A. For each of the potential positions around the TSS, we aggregate the value for each of the known genes. For each pair of positions, this results in two vectors which are then correlated using a spearman rank order correlation. In our example, we would aggregate all the values at the transcription start sites for the $H_{3}K_{27}ac$ signal (first red line) and for the $Pol-II$ signal (first blue line). The value obtained after correlation is then placed into the correlation graph at the TSS. The same process would be repeated for position $TSS+x$, with $x$ ranging from $\left[-1000,1000\right]$ (second red and blue line). Figure 3A shows the direct correlation between the TSS and surrounding area between $H_{3}K_{27}ac$ and $Pol-II$.

Figure 3: If we have two collections of transcription start sites and surrounding areas, one for instance for H3K27ac and another for Pol-II, then it becomes possible to correlate each position within these vectors and report how well a signal in H3K27ac correlates with a signal in Pol-II. A) The correlation itself is calculated by aggregating the values at the transcription start site for the two vectors and then correlating these two vectors using a spearman rank order correlation. The resulting number is set out in the graph at the position from which the data points where taken. This process can be repeated for each position in the neighborhood of the transcription start site. B) To check whether a signal in H327ac correlates with a signal more downstream in Pol-II we follow a similar pattern, but instead of aggregating the data points at the same position we will now aggregate the signal from the H3K27ac track at the TSS, while we obtain the data from the Pol-II signal at TSS-1000. If we repeat this over the entire static track then we can draw again a correlation graph for each position in the static signal. C) To lay out all the correlations with and without shift we convert each correlation graph to a line filled with hue values. Each line is then placed at the appropriate position in the 2D correlation map.

This process would then also be repeated for each possible combination of $TSS+x$ against every other $TSS+y$. Figure 3B illustrates the correlation between $H_{3}K_{27}ac$ at the $TSS$ (first red line) against the $Pol-II$ signal at $TSS-1000$ (first blue line).

Because we have around 2000 correlation values for each potential translations, which amount to another 4000 positions), we need a good manner to visualize the results. This is achieved by converting each correlation profile to a line in which the x-coordinate reflects the position in the static track and the hue (the color) reflects the strength of the correlation. Each of these lines is then placed in a 2 dimensional map in which the height reflects a translation of the dynamic track. The middle of the map reflects no shift between the static and dynamic signal. If we go up, the dynamic signal was shifted to the left, if we go down the dynamic signal was shifted to the right. Figure 3C presents this idea. The intensity of a pixel in the correlation map reflects the variance of the two signals. If we find a vertical dark line it means that the static signal had a very low variance around this area. A vertical slanted line indicates a low variance in the dynamic track at that specific position.

A strong correlation at position $\left(x,y\right)$ in those correlation maps indicates that the behavior at position $TSS+x$ (signal versus no signal) will be similar to the behavior at $TSS+y$ in the second track. In figure 3C the maximum correlation appears at $\left(480,50\right)$ between a $H_{3}K_{27}ac$ track (fixed) and a $Pol-II$ track (dynamic). This indicates that the $H_{3}K_{27}ac$ signal at $TSS+480$ will correlate strongly ($r=0.73$) with the with $Pol-II$ signal at $TSS+50$. As can be observer, one interesting aspect of these plots is that we can directly answer which positions will strongly relate to each other, even when it involves a translation of one the signaling tracks.

3.2 Patterns

Figure 4A: Each type of these figures explores the correlation between two signals $A$ (horizontal, pictured in red) and $B$ (vertical, pictured in blue) by visualizing the area in each of the signals and then superimposing them (the purple graph).A) Correlations that appear as a point, disc or ellipse in the correlation map at position $\left(x,y\right)$ imply that if there is a strong signal $A$ at $TSS+x$, then there will also tend to be a strong signal $B$ at $TSS+y$. The width of the ellipse reflects the width of the $A$ signal over which this correlation holds. The height of the ellipse reflects the width of the $B$ signal over which this correlation holds true.

When measuring the correlations we do not only find single unique strong correlations. In general correlations will co-localize and appear as larger discs or ellipses, which directly indicates the size of the peak over which correlation could be measured. Figure 4A illustrates this. We find a strong correlation for signal $A$ at $TSS-100$ with signal $B$ at $TSS+150$. The width of the ellipse is around 50 bases and its height is 200 bases. This indicates that a peak of around 50 bases wide in the $A$ track correlates with a 200 bases wide peak in track $B$.

Figure 4: Each subfigure explores the correlation between two signals $A$ (horizontal, pictured in red) and $B$ (vertical, pictured in blue) by visualizing the area in each of the signals and then superimposing them (the purple graph).A) Correlations that appear as a point, disc or ellipse in the correlation map at position $\left(x,y\right)$ imply that if there is a strong signal $A$ at $TSS+x$, then there will also tend to be a strong signal $B$ at $TSS+y$. The width of the ellipse reflects the width of the $A$ signal over which this correlation holds. The height of the ellipse reflects the width of the $B$ signal over which this correlation holds true. The correlation maps have, aside from discs and ellipses, some other recurring patterns as well. BCD) These are vertical, horizontal and diagonal lines. Each pattern has a specific meaning and can reflect a correlation between a peak and an area, between 2 areas or between an area and a peak. For instance, figure C shows that the $A$ signal from $TSS-400$ to $TSS-200$ (red graph) will correlate with the $B$ signal at $TSS+250$. This combination is shown in the combination plot.

Aside from the expected correlations between peaks, other patterns will emerge in the correlation maps as well. These are horizontal lines, vertical lines and diagonal lines. We discuss each of the cases below because each of them has a specific and important meaning.

To understand what a vertical line means (Figure 4B) it is best to understand what the two endpoints of such a line mean. Afterward we can interpolate between them and get a better understanding of the full picture. Let us assume for the sake of argument that we have a vertical line from $\left(-200,100\right)$ to $\left(-200,250\right)$ in a correlation map between $A$ (x-axis) and $B$ (y-axis). The first endpoint tells us that signal $A$ at $TSS-200$ will correlate with signal $B$ at $TSS+100$. The second endpoint tells us that signal $A$ at $TSS-200$ will correlate with signal $B$ at $TSS+250$. Interpolating between these two extremes reveals that $A$ at $TSS-200$ will correlate with $B$ ranging from $TSS+100$ to $TSS+250$.

A horizontal line in the correlation maps (Figure 4C) has a similar meaning although in this case signal $B$ will be localized, while signal $A$ will cover an entire area. Assume that we have a line from $\left(-400,250\right)$ to $\left(-200,250\right)$. At the first endpoint we know that signal $A$ at $TSS-400$ correlates with the signal $B$ at $TSS+250$. The second endpoint indicates that signal $A$ at $TSS-200$ correlates with signal $B$ at $TSS+250$. Interpolating between these two extremes shows that we will have a correlation between $B$ at $TSS+250$ and $A$ ranging from $TSS-400$ to $TSS-200$. We find that vertical or horizontal lines indicates a correlation between a local element (peaks for instance) and a larger element (such as an entire area).

Diagonal lines (Figure 4D) is the last case that commonly occurs and indicates that two large areas directly correlate against each other. Assume a line from $\left(-100,-100\right)$ to $\left(250,250\right)$. The first point implies that signal $A$ and $B$ correlate at $TSS-100$. The second endpoint implies that signal $A$ and $B$ correlate at $TSS+250$. Interpolating between these two leads to a correlation between signal $A$ (range $TSS-100$ to $TSS+250$) and signal $B$ (range $TSS-100$ to $TSS+250$).

In general diagonal lines will bloom out a bit (Figure 7A) because of a transitive closure between three correlations. When signal $A$ and $B$ correlate at $TSS+x$, at $TSS+y$ and $TSS+z$, then so will signal $A$ at $TSS+x$ and $TSS+y$ because the behavior of the signal at points $x$, $y$ and $z$ will very likely be similar as well. If the blooming is small then one can assume there is a specific positional effect driving the correlations. The blooming effect is larger for diagonal lines that do not go through the origin. If the diagonal correlations are very localized one might also start thinking about a sequence specific bias and we did indeed find these. We come back to them in section 4.1.

At the moment we know that the correlation maps can exhibit patterns which make it possible to find relations between peaks and peaks (discs or ellipses), peaks and areas (vertical and horizontal lines) and between areas and areas (large correlation patches).

Figure 5: The correlation maps often exhibit recurring patterns. These are A) local spots, indicating a correlation between the signals at the specified position. In this example the $CTCF$ at $TSS-50$ correlates with the $Pol-II$ signal at $TSS-50$. The width of this correlation is around 100 base pairs. B) Another typical pattern are diagonal correlation through or almost through the 0 point. These indicate that both signals correlate well towards each other, regardless of the respective position around the TSS. In this case $H_{2}BK_{5}me_{1}$ correlates with $H_{3}R_{2}me_{1}$ with $r=0.2$. C) Vertical correlations indicate that the signal at one position ( $H_{3}K_{79}me_{1}$ at intron 1, exon 2 and exon 3) correlates with a larger area of the second signal ($H_{3}K_{27}ac$ from $TSS-600$ to $TSS-150$ and the first exon). D) Horizontal correlations indicate a similar pattern, in which an area of the first signal correlates with a well defined position in the second signal. E.g; the $H_{2}BK_{5}ac$ area from $TSS-600$ to $TSS-150$ and the first exon correlates with the $H_{3}K_{36}me_{3}$ signal in exon 4 to exon 10.

H3R2me1 ~ H2BK5me1
H3K79me1 ~ H3K27ac
H2BK5ac ~ H3K36me3

To demonstrate some of these patterns we selected 4 correlation maps (Figure 5). The first one shows a localized correlation between $CTCF$ and $Pol-II$ (Figure 5A). In this example the $CTCF$ at $TSS-50$ correlates with the $Pol-II$ signal at $TSS-50$. The width of this correlation is around 100 base pairs. Figure 5B shows a diagonal correlation pattern through the $TSS$ for $H_{3}R_{2}me_{1}$ and $H_{2}BK_{5}me_{1}$. These indicate that the both signals correlate well with each other, regardless of the respective position around the $TSS$. In this case $H_{2}BK_{5}me_{1}$ correlates with $H_{3}R_{2}me_{1}$ with $r=0.2$. Figure 5C demonstrates a number of vertical correlations between $H_{3}K_{79}me_{1}$and $H_{3}K_{27}ac$. The signal at $H_{3}K_{79}me_{1}$, intron 1, exon 2 and exon 3 correlate with a signal of $H_{3}K_{27}ac$ in front of the $TSS$ (from $TSS-600$ to $TSS-150$). Interesting here is also that the $H_{3}K_{79}me_{1}$signal at intron1, exon2 and exon3 correlate with the $H_{3}K_{27}ac$ signal in the first exon1. Figure 5D shows a number of horizontal correlation patterns between $H_{2}BK_{5}ac$ and $H_{3}K_{36}me_{3}$. Here, the $H_{2}BK_{5}ac$ area from $TSS-600$ to $TSS-150$ and the first exon correlates with the $H_{3}K_{36}me_{3}$ signal in exon 4 to exon 10.

3.3 Correlation Mining

Using the histone methylation and acetylation patterns, we calculated in total around $16.10^{10}$ correlations, a number sufficiently large to start thinking about finding important correlations automatically. The tool we developed will take from a collection of correlation maps the strongest ones and present them in a combined map (called the maximized map). The color in the maximized map no longer reflects the strength of a correlation, this is now shown through the the pixel luminosity. The color itself is now used to reflect the map with the strongest correlation at that particular position. By then reading the color one can find out which correlation map was the reason for a large correlation. The colors are chosen such that neighboring correlation areas will have a high color contrast.


Figure 6: Maximized correlation maps. Each figure contains around 40 different correlations maps. The color reflects which one. The intensity of a pixel reflects the correlation strength. A) Horizontal CTCF is placed, vertical any of the 41 other correlation maps is visualized. The 4 most prominent colors are red: $H_{4}K_{20}me_{1}\left(r=0.39\right)$, green: $H_{3}K_{36}me_{3}\left(r=0.40\right)$, turquoise: $PolII\left(r=0.49\right)$ and purple: $H_{3}K_{4}me_{3}\left(r=0.55\right)$. Interesting on this maximized plot is that the right half of the picture is dark, indicating that there never is a correlation between a $CTCF$ signal in the gene body and any other methylation and/or acetylation signal. In plot B and C we looked for the maximal correlations against $H_{3}K_{4}me_{3}$. In plot B we find that $H_{3}K_{36}me3$, $H_{4}K_{20}me1$, $H_{3}K_{9}me_{1}$, $PolII$ and $H_{3}K_{18}ac$ are relevant. After removing these correlations, we created a new maximized map. C) The 2nd layer maximized map shows a number of extra correlations. In particular $H_{3}K_{27}ac$, $H_{2}AZ$, $H_{2}BK_{120}ac$, $H_{4}K_{91}ac$ and $H_{3}K_{79}me_{1}$. To understand how each correlation in itself is distributed, one can fall back to the original correlation maps (see Figure 7)

B & C

Figure 6 presents three maximized correlation maps. Figure 6A shows the maximized output for all the maps involving $CTCF$ horizontally. What directly becomes apparent is that none of the methylations patterns relate in any significant manner to the CTCF track in the gene body. In front of the $TSS$ however we find that certain areas correlate sufficient towards $CTCF$. To read these plots one can start for instance at the TSS and assume that we have a CTCF signal there, then one start 100 bases in front of the $TSS$, which is in the map 100 bases down. There we find $Pol-II$, if we now go into the gene body (that is up on the plot) we find that when there is a CTCF signal at the TSS, we will find in succession the following strong correlations: $H_{4}K_{20}me_{1}$ up to the 4th exon and then $H_{3}K_{36}me_{3}$ from exon4 on.

Of course, since we only show the maximal correlations, certain areas can be masked out and not visible despite that they also have a strong correlations. To make these visible we can retrieve multiple layers of the correlation map. The first layers includes all correlation maps. The second layer only includes the correlation maps that where not sufficiently present in any of the previous maps. If we do this for $H_{3}K_{4}me_{3}$ (Figure 6B and 6C) we find that a strong $H_{3}K_{4}me_{3}$ signal in front of the $TSS$ relates to the following other histone modifications:

  1. From exon 4 to exon 10: $H_{3}K_{36}me_{3}$ ($r=0.47$, first layer) and $H_{3}K_{27}me_{1}$ ($r=0.33,$ 2nd layer)
  2. From exon 2 to intron 3: $H_{4}K_{20}me_{1}$ ($r=0.46$, 1st layer) and $H_{3}K_{79}me_{1}$ ($r=0.4$, 2nd layer)
  3. Up to exon 2: $H_{3}K_{9}me_{1}$ ($r=0.64$, 1st layer)
  4. From $TSS-200$ to $TSS$: 1st layer: $Pol-II$ with $r=0.79$. The second layer starts to make a distinction between the $H_{3}K_{4}me_{3}$ signal position. If it is from $TSS-1000$ to $TSS-500$ we find the strongest correlation towards $H_{4}K_{91}ac$ ($r=0.75$). From $TSS-500$ to $TSS$ we find the strongest correlation towards $H_{3}K_{27}ac$ ($r=0.8$).
These distinct correlation areas can then be found back in the original correlation maps, as is shown in Figure 7.

Figure 7: A variety of $H_{3}K_{4}me_{3}$ correlation maps. The selection of correlation maps is based on the maximized maps from figure 6. $H_{3}K_{4}me_{3}$ is in each map placed on the horizontal axis. the vertical axes are as follows. Top) $H_{2}AZ$, $H_{2}BK_{120}ac$, $H_{3}K_{9}me_{1}$ and $H_{3}K_{18}ac$. Middle) $H_{3}K_{27}ac$, $H_{3}K_{27}me_{1}$, $H_{3}K_{36}me_{1}$ and $H_{3}K_{79}me1$. Bottom) $H_{4}K_{19}ac$, $Pol-II$ and $H_{4}K_{20}me_{1}$.

3.3.1 Significantly altered correlations

A second technique employed by the maximizer is the possibility to show only significantly altered correlations. The rationale behind this was that in one of our initial experiments we found that the upstream region would always exhibit strong correlations. In the end it turned out that the signal mainly reflected whether an exon or intron was present and this 'hidden' factor would drive the correlation upwards. Although we need to come back to these hidden factors in section 4.3, we tested an approach in which we would use all 41 correlation maps and assess what the expected correlation was at each position. Once we had the expected correlation it was not difficult to only report the most significant altered correlation. This approach is somewhat adhoc because we no longer know how strong the correlation, barring all hidden factors, really is.

4 Data Analysis Results & Discussion

In this section we will further bring together the different approaches we tried and report on which techniques worked and which should be avoided, including argumentation

As an initial test of our algorithm we created a random track in which we placed 10000 random peaks and correlated these against one of the signal tracks (Figure 8A). As a $2^{\mbox{nd}}$ test we correlated a signal track against a track that contained 100'000'000 random alignments (Figure 8B). In none of the cases did we find a correlation that was worrisome. They were all below the correlation strength that one expects with 20000 transcription start sites (~0.03)

Figure 8: As an initial validation of our algorithm we created a A) random track in which we placed 10000 random peaks and correlated these against the $H_{3}K_{27}me_{3}$ track. As a $2^{\mbox{nd}}$ test we B) correlated a track containing 100'000'000 random alignments against $H_{3}K_{27}me_{1}$. In neither cases did we find a correlation that was worrisome. They were all below the correlation strength that one expects with 20000 transcription start sites (~0.03). As a test of the GC-bias hypothesis we generated A) a signal reflecting the GC content of the genome (horizontally) and vertically $H_{3}K_{27}me_{1}$ was set out. The maximal correlation was 0.03 as well.
A (Random peaks)
B (Noise)
C (GC Content)

4.1 Fragment Extension

We first report on the difference between relying on only sequenced bases (extend alignment positions up to the read-length) against the bases present in the sample (extend alignments up to the fragment length).

If we do not extend the fragments we find that the alignment will mainly measure the areas where the fragmentation process choose to break up DNA strands. This is confirmed by pointing out that the maximized map of methylation correlations only shows strong correlations at the diagonal. This indicates a strong positional bias and maybe even a lighter form of sequencing bias. In this case, the ultra high resolution of the sequencing technology is somewhat of a disadvantage because most researchers are not interested in understanding where the fragments got broken up. They merely want to know whether a specific base was in the sample.

When we extent each aligned fragment to 160 bases we find stronger correlations, which indicates that the non-extended interpretation has a certain amount of sampling noise. Secondly, the extended alignment lead to correlation maps that do not only have strong correlations near the diagonal. Both make for a compelling argument to perform a smoothing on the data. If one wants to use deep sequencing data to study CHIP-SEQ data, it is necessary to extend all the alignment position up to the expected fragment length. If one is interested in sequencing errors and positional biases then of course this should not be done.

Because a positional bias might have been related to the GC content of the genome we correlated the GC content of a random track against its measured signal and we did not find any significant correlation (Figure 8C).

4.2 Data Normalization

Up to now we did not go into much detail regarding the two normalization schemes we used (Fig. 1B). The idea was that one can of course use the signal as measured. However, since we compare them against each other it might have been useful to make some form of internal control such that each transcription area is normalized internally. The rationale behind this was that certain phenomena could affect chip binding or transcription at a kilobase resolution: e.g: is this part of the chromosome close to the centromere, is it accessible etcetera. The second reason we considered normalization was that, if we would want to create a predictor of transcription start sites based on the measurements it is necessary to be able to normalize each vector individually; independent of some peculiarities that might have affected the signal strength. As a third reason we considered that the shape of some of some curves could be related to transcription start sites and not the absolute signal.

After doing some testes we found that normalization offers little quantitative difference with respect to the maximal correlations. There is however a qualitative difference which can be best understood by looking at how curve shape, normalization and signal strength intertwine.

In the first case (Fig. 1B) where we only use the measured value we find that transcript areas with strongest signal will tend to dominate the final profile. This in itself is not such a good thing because it leads to the wrong impression that the shape of the average signal is the average shape, which it in reality is not. The final shape can be driven by some outliers. This problem of outliers however is limited in the case of the correlator because the signals are ranked before they are fed into a linear pearson correlator. This ranking process aside, the correlator will still not be able to recognize specific shapes in the signal.

In the second case, when we normalize the data, we find that if there is a common shape present this shape will be preserved in the final profile. In contrast to the previous absolute interpretation, not a single vector can now dominate the total profile. It is however still possible that a large group of similarly behaving genes (inactive genes) altogether affect the outcome due to an amplification of 'no signal'.

When used in a correlation context (based on spearman rank order correlation) it seems most appropriate to not normalize the data. However, when creating profiles, it seems more appropriate to use both approaches and observer the differences between them.

4.3 Multi-hits

Figure 9: The effect preprocessing of the data has on the correlation strengths is dramatic. Row 1 contains the cross correlations between $H_{3}K_{23}ac\sim H_{3}K_{27}me_{3}$, row 2 contains the cross correlations between $H_{3}K_{79}me_{2}\sim H_{3}K_{27}me_{3}$. In column 1 we counted each multi-hit as if they were multiple single hits (w=1). In the second columns we weighed each multi-hit depending on the number of alignment positions ( $w=1\backslash n$). In the third column we did not count multi-hits and thus use a weight of 0 ($w=0$). In the fourth column, we introduced a new approach where we marked each of the positions in the genome subject to a multi-hit as 'unknown'. The correlation strengths are from left to right and top to bottom: 0.52, 0.17, 0.12, 0.055; 0.54, 0.12, 0.056, 0.017. As can be observed, the effect on the correlation strength can be dramatic, which indicates that most commonly used techniques to deal with multi-hits have a strong bias towards strong positive correlations.

One of the most important observations from the preliminary results was that all correlations were positive in the upstream region around 1000 bases and within the intron areas. A closer investigation revealed that although many genes had no signal in the introns, some of them did so in abundance in all tracks. This made us reexamine the assumption that we could treat multi-hits as repeated single hits and we did a number of small experiments to verify the impact multi-hits had on the correlation values. Figure 9 illustrates some of these results.

In most cases, the correlation values would be around 0.52 if we counted each hit individually. If we weighed each multi-hit the correlations lowered to a maximum of 0.17, and when we relied only on the uniquely matching hits (weight=0), the correlations lowered to 0.12. In another case we found that counting each multi hit provided us with a correlation of 0.63, weighing them gave 0.27 and only counting unique ones gave a value of 0.13.

Because the treatment of multi-hits had such a high impact on the correlation and because none of the three approaches are scientifically correct, we started to think about a better strategy to deal with multi-hits, which would preserver the measured information correctly. For repeat areas we don't know how many times a specific base was present in the sample, as such we marked these areas as 'unknown' and passed this information to the correlator which would then treat these points as missing values. This precise interpretation led to an even more dramatic lowering of the correlations: to 0.055 in case of $H_{3}K_{23}ac\sim H_{3}K_{27}me_{3}$ and to 0.017 for $H_{3}K_{79}me_{2}\sim H_{3}K_{27}me_{3}$. When using the precise interpretation of the data we also no longer found the intron correlations. This indicates that none of the commonly used methods to deal with multiple alignments is appropriate when a subsequent analysis step involves correlations.

4.4 Exon/Intron correlations

The realization that we sometimes find different correlation strengths between exons and introns might indicate that it could be possible to detect whether something is an exon or intron solely based on the histone methylation/acetylation patterns. This might be possible, however one should carefully note a number of limitations of using these correlations in a predictive context. The first limitation is that we rank the data set so we cannot directly go from the measured signal to a rank independent of knowing where the other exons/introns are. Of course, the spearman rank order correlation is often a more stringent measure than the linear pearson correlation, we might assume that the LP correlation could be higher and this ranking problem might not really be a problem.

The second observation is that the scale of the images is dynamically determined based on the maximum correlation and none of the correlations are of such high value that we can readily jump from an apparent pattern to a predictor. That is: a correlation of 0.58 in Figure 5D and a correlation of 0.41 in Figure 5C do not comfortably suggest that it might be possible to use the data directly as a univariate predictor.

The third observation is that we not only need to look at the correlation strength in the exons (which is low), but also whether we will be able to distinguish between the correlations in the introns. If we look at the difference between the exons and introns in correlation strength the picture becomes even more bleak: 0.0735 for 5D and 0.052 for 5C. Obviously it will be a pretty hard task to use such correlations to distinguish between exons and introns. This also illustrates that the correlation maps should be read with care. The hue color scheme is optimal to allow us to recognize specific areas, but the actual correlation values should not be omitted in an interpretation.

5 Conclusion

In this article we presented a novel methodology to calculate correlations between genomic signal tracks. For instance, one can correlate each position around the TSS of one signal track against the signal at another position around the TSS of a second signal track. We observed that the resulting correlation maps tend to be composed of discs, ellipses, horizontal, and vertical lines. Such recurring patterns can be interpreted as correlations between peak/peaks, peaks/areas or areas/areas.

In order to sort through 21000 of these correlation maps we created an automatic correlation finder that presents layers of maximal correlations in a coherent fashion.

We furthermore found that repeat areas, and more specifically, hits with multiple alignment positions should be treated as missing data. None of the three existing options nowadays: (not counting, weighing, treating each of them as an individual) respects the measured data and this leads in almost all cases to higher correlations. By treating missing data as such, we saw correlations lower from 0.52, to 0.17, to 0.12 to 0.05 in one case and from 0.63, to 0.27, to 0.13 to 0.017 in another case.

Extending aligned fragments to the selected fragment size of, for instance 160 bases, improves the correlations. Without fragment extension the data seem to be positionally biased.

Transcription areas can be or cannot be normalized. There is mainly a qualitative impact. With normalization the correlations might lower due to amplification of noise, the shape of the profiles is however taken into account. Without normalization, the shape of the curves is not taken into account.

We also found that the GC content of the genome does not affect the measurement sufficiently to be of concern. The observation that most of the patterns correlate across longer stretches indicates that the widely perceived model of peaks is premature and although histones might bind at specific positions, this is certainly not the only interpretation that one should give to the data.


I would like to thank Christian Beissel for introducing me to the field of epigenetics, a field which is otherwise unknown to me. And of course Philipp Bucher, who has shown great interest in this article. If it where not due to administrative constraints, we would probably be cooperating on this research.


1.High-resolution profiling of histone methylations in the human genome. Artem Barski, Suresh Cuddapah, Kairong Cui, Tae-Young Roh, Dustin, E. Schones, Zibhin Wang, Gang Wei, Louri Chepelev, and Keji Zhao Cell, 129:823-837, 2007
2.Combinatorial patterns of histone acetylations and methylations in the human genome Zhibin Wang, Chongzhi Zang, Jeffrey A. Rosenfeld, Dustin E. Schones, Artem Barski, Suresh Cuddapah, Kairong Cui, Tae-Young Roh, Weiqun Peng, Michael Q.Zhang, and Keji Zhao Nature Genetics, 40(7), July 2008
3.The NCBI Short Read Archive http://www.ncbi.nlm.nih.gov/Traces/sra/sra.cgi
4.Ultra high throughput alignment of short sequence tags Anthony J. Cox Declared to be 'in preparation' according to Illumina; 2003
5.Eland2Wig: A Program to convert Eland aligned output to Wig files Werner Van Belle http://analysis.yellowcouch.org/programs/Eland2Wig