Limits...
Inherent signals in sequencing-based Chromatin-ImmunoPrecipitation control libraries.

Vega VB, Cheung E, Palanisamy N, Sung WK - PLoS ONE (2009)

Bottom Line: The growth of sequencing-based Chromatin Immuno-Precipitation studies call for a more in-depth understanding of the nature of the technology and of the resultant data to reduce false positives and false negatives.We found that copy number plays a major influence in both ChIP-enriched as well as control libraries.Next, significantly tag-rich 5 kbp regions were identified and they were associated with various genomic landmarks.

View Article: PubMed Central - PubMed

Affiliation: Computational and Mathematical Biology Group, Genome Institute of Singapore, Singapore, Singapore.

ABSTRACT

Background: The growth of sequencing-based Chromatin Immuno-Precipitation studies call for a more in-depth understanding of the nature of the technology and of the resultant data to reduce false positives and false negatives. Control libraries are typically constructed to complement such studies in order to mitigate the effect of systematic biases that might be present in the data. In this study, we explored multiple control libraries to obtain better understanding of what they truly represent.

Methodology: First, we analyzed the genome-wide profiles of various sequencing-based libraries at a low resolution of 1 Mbp, and compared them with each other as well as against aCGH data. We found that copy number plays a major influence in both ChIP-enriched as well as control libraries. Following that, we inspected the repeat regions to assess the extent of mapping bias. Next, significantly tag-rich 5 kbp regions were identified and they were associated with various genomic landmarks. For instance, we discovered that gene boundaries were surprisingly enriched with sequenced tags. Further, profiles between different cell types were noticeably distinct although the cell types were somewhat related and similar.

Conclusions: We found that control libraries bear traces of systematic biases. The biases can be attributed to genomic copy number, inherent sequencing bias, plausible mapping ambiguity, and cell-type specific chromatin structure. Our results suggest careful analysis of control libraries can reveal promising biological insights.

Show MeSH

Related in: MedlinePlus

Comparison of genome-wide copy number from three mouse cell types (ES, NP, and MEF), sorted in chromosomal order.Although copy number wise, they were highly similar (Pearson's r>0.74 for all pairings) as expected, the exclusively high correlation (Pearson's r = 0.946) between ES and NP reflected their relationship at sample preparation level [8], [9].
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2666154&req=5

pone-0005241-g002: Comparison of genome-wide copy number from three mouse cell types (ES, NP, and MEF), sorted in chromosomal order.Although copy number wise, they were highly similar (Pearson's r>0.74 for all pairings) as expected, the exclusively high correlation (Pearson's r = 0.946) between ES and NP reflected their relationship at sample preparation level [8], [9].

Mentions: Similar analyses were also performed using three mouse WCEseq libraries published by Mikkelsen et al. [8] which were generated from embryonic stem (ES), neural progenitor (NP), and embryonic fibroblasts (MEF) cells. Although the copy number estimates across these three libraries are generally similar (Pearson's r>0.74 for all pairings, Table 1), some differences were still apparent (Fig. 2). The correlation between that of ES and NP was unexpectedly high at almost 0.95, while the correlation between MEF and the other two libraries was about 0.75 on average. Although the copy numbers of these three cell types are expected to be very similar, the perceptible difference could be due to other reasons. One potential explanation could be due to how the libraries were generated. For example, the NP cells were derived from the ES, while the MEF was obtained independently [8], [9].


Inherent signals in sequencing-based Chromatin-ImmunoPrecipitation control libraries.

Vega VB, Cheung E, Palanisamy N, Sung WK - PLoS ONE (2009)

Comparison of genome-wide copy number from three mouse cell types (ES, NP, and MEF), sorted in chromosomal order.Although copy number wise, they were highly similar (Pearson's r>0.74 for all pairings) as expected, the exclusively high correlation (Pearson's r = 0.946) between ES and NP reflected their relationship at sample preparation level [8], [9].
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2666154&req=5

pone-0005241-g002: Comparison of genome-wide copy number from three mouse cell types (ES, NP, and MEF), sorted in chromosomal order.Although copy number wise, they were highly similar (Pearson's r>0.74 for all pairings) as expected, the exclusively high correlation (Pearson's r = 0.946) between ES and NP reflected their relationship at sample preparation level [8], [9].
Mentions: Similar analyses were also performed using three mouse WCEseq libraries published by Mikkelsen et al. [8] which were generated from embryonic stem (ES), neural progenitor (NP), and embryonic fibroblasts (MEF) cells. Although the copy number estimates across these three libraries are generally similar (Pearson's r>0.74 for all pairings, Table 1), some differences were still apparent (Fig. 2). The correlation between that of ES and NP was unexpectedly high at almost 0.95, while the correlation between MEF and the other two libraries was about 0.75 on average. Although the copy numbers of these three cell types are expected to be very similar, the perceptible difference could be due to other reasons. One potential explanation could be due to how the libraries were generated. For example, the NP cells were derived from the ES, while the MEF was obtained independently [8], [9].

Bottom Line: The growth of sequencing-based Chromatin Immuno-Precipitation studies call for a more in-depth understanding of the nature of the technology and of the resultant data to reduce false positives and false negatives.We found that copy number plays a major influence in both ChIP-enriched as well as control libraries.Next, significantly tag-rich 5 kbp regions were identified and they were associated with various genomic landmarks.

View Article: PubMed Central - PubMed

Affiliation: Computational and Mathematical Biology Group, Genome Institute of Singapore, Singapore, Singapore.

ABSTRACT

Background: The growth of sequencing-based Chromatin Immuno-Precipitation studies call for a more in-depth understanding of the nature of the technology and of the resultant data to reduce false positives and false negatives. Control libraries are typically constructed to complement such studies in order to mitigate the effect of systematic biases that might be present in the data. In this study, we explored multiple control libraries to obtain better understanding of what they truly represent.

Methodology: First, we analyzed the genome-wide profiles of various sequencing-based libraries at a low resolution of 1 Mbp, and compared them with each other as well as against aCGH data. We found that copy number plays a major influence in both ChIP-enriched as well as control libraries. Following that, we inspected the repeat regions to assess the extent of mapping bias. Next, significantly tag-rich 5 kbp regions were identified and they were associated with various genomic landmarks. For instance, we discovered that gene boundaries were surprisingly enriched with sequenced tags. Further, profiles between different cell types were noticeably distinct although the cell types were somewhat related and similar.

Conclusions: We found that control libraries bear traces of systematic biases. The biases can be attributed to genomic copy number, inherent sequencing bias, plausible mapping ambiguity, and cell-type specific chromatin structure. Our results suggest careful analysis of control libraries can reveal promising biological insights.

Show MeSH
Related in: MedlinePlus