 Software
 Open Access
 Published:
MotifAdjuster: a tool for computational reassessment of transcription factor binding site annotations
Genome Biology volumeÂ 10, ArticleÂ number:Â R46 (2009)
Abstract
Valuable bindingsite annotation data are stored in databases. However, several types of errors can, and do, occur in the process of manually incorporating annotation data from the scientific literature into these databases. Here, we introduce MotifAdjuster http://dig.ipkgatersleben.de/MotifAdjuster.html, a tool that helps to detect these errors, and we demonstrate its efficacy on public data sets.
Rationale
The regulation of gene expression involves a complex system of interacting components in all living organisms [1] and is of fundamental interest, for instance, for cell maintenance and development. One level of regulation is realized by DNAbinding transcription factors (TFs). The DNAbinding domain of a TF is capable of recognizing specific binding sites (BSs) in the promoter regions of its target genes [2]. Binding of a TF can induce (activator) or inhibit (repressor) the transcription of its target genes. The general ability to control a target gene may depend on the BS itself, its strand orientation, and its position with respect to the transcription start site. If other BSs are present, the ability of a TF to bind the DNA may additionally depend on strand orientations and positions of these BSs.
One important prerequisite for research on gene regulation is the reliable annotation of BSs. The approximate regions on the doublestranded DNA sequence bound by TFs can be determined by wetlab experiments such as electrophoretic mobility shift assays (EMSAs) [3], DNAse footprinting [4], enzymelinked immunosorbent assay (ELISA) [5, 6], ChIPchip [7], or mutations of the putative BS and subsequent expression studies. Because TFs bind to doublestranded DNA, the strand annotations of nonpalindromic BSs in the databases are either missing or added, based on manual inspection or predictions from bioinformatics tools such as MEME [8], Gibbs Sampler [9, 10], Improbizer [11], SeSiMCMC [12], or AGLAM [13].
After wetlab identification, data about transcriptional gene regulatory interactions, including the annotated BSs, are published in the scientific literature. Subsequently, these data are extracted by curation teams and manually entered into databases on transcriptional gene regulation such as CoryneRegNet [14], PRODORIC [15], or RegulonDB [16] for prokaryotes, and AGRIS [17], AthaMap [18], CTCFBSDB [19], JASPAR [20], OregAnno [21], SCPD [22], TRANSFAC [23], TRED [24], or TRRD [25] for eukaryotes. Three typical problems may occur during the process of transferring these data.
First, erroneously annotated BS: This error may occur in the original study or during the transfer process from the scientific literature to the databases. A sequence is declared to contain a BS, although, in reality, it does not.
Second, shift of the BS: The BS may be erroneously shifted by one or a few base pairs. This typically happens during the transfer process from the scientific literature to the databases.
Third, missing or wrong strand orientation of the BS: The strand orientation of a BS is often not or incorrectly annotated. For example, all BS orientations are arbitrarily declared to be in 5'â†’3' direction relative to the target gene in CoryneRegNet and in RegulonDB [14, 16].
These problems can strongly affect any of the subsequent analysis steps, such as the inference of sequence motifs from "experimentally verified" data, the calculation of P values for the occurrence of BSs, the detection of putative BSs in genomewide scans and their experimental validation, or the reconstruction of transcriptional generegulatory networks.
Here, we introduce MotifAdjuster, a software tool for detecting potential BS annotation errors and for proposing possible corrections. Existing bioinformatics tools [8â€“13] are not optimized for this task (Additional data file 1), because they do not allow shifting the BS by using a nonuniform distribution and considering both strands with unequal weights. In contrast, MotifAdjuster allows the user to incorporate prior knowledge about (i) the probability of erroneously annotated BSs, (ii) the distribution of possible shifts, and (iii) the strand preference.
One widelyused model for the representation of BSs is the position weight matrix (PWM) model [8â€“13, 26, 27], and many software tools for genomewide scans of sequence motifs are based on PWM models [26, 28, 29]. MotifAdjuster is based on a simple mixture model using a PWM model on both strands for the motif sequences and a homogeneous Markov model of order 0 for the flanking sequences similar to MEME, Gibbs Sampler, Improbizer, SeSiMCMC, or AGLAM. For a given set of BSs, MotifAdjuster tests whether each sequence contains a BS, and it refines the annotations of position and strand for each BS, if necessary, by maximizing the posterior of the mixture model by using a simple expectation maximization (EM) algorithm.
To test the efficacy of MotifAdjuster, we apply it to seven data sets from CoryneRegNet, and we record for each of them the set of potential annotation errors. For one example, the nitrate regulator NarL, we compare the proposed adjustments with the original literature, with a manual strand reannotation of the BS strands, and with an independent and handcurated reannotation provided by PRODORIC. Finally, we test whether the PWM estimated from the adjusted NarL BSs can help to detect unknown BSs in those promoter regions that are known to be bound by NarL, but for which no BS could be predicted in the past.
Algorithm
In this section, we present the MotifAdjuster algorithm including the mixture model, the prior, and the maximum a posteriori (MAP) estimation of the model parameters given the data.
Mixture model
We denote a DNA sequence of length L by x:= (x_{1}, x_{2}, ..., x_{ L }), the nucleotide at position â„“ âˆˆ [1, L] by x_{ â„“ }âˆˆ {A, C, G, T}, and the reverse complement of x by x^{RC}. For modeling a BS x of length w, we use a PWM model, which assumes that the nucleotides at all positions are statistically independent of each other, resulting in an additive loglikelihood
of sequence x given the model parameters Î» [30, 31], where the subscript f stands for foreground. Here, denotes the logarithm of the probability of finding nucleotide a âˆˆ {A, C, G, T} at position â„“, Î»^{â„“}denotes the fourdimensional vector , and Î» denotes the (4 Ã— w) matrix, that is, Î» denotes the PWM [32â€“36].
For modeling the flanking sequences, we use a homogeneous Markov model of order 0, which assumes that all nucleotides are statistically independent, resulting in an additive loglikelihood
of sequence x given model parameters Ï„ [32â€“36], where the subscript b stands for background. Here, Ï„_{ a }denotes the logarithm of the probability of nucleotide a, and Ï„ denotes the vector (Ï„_{ A }, ..., Ï„_{ T })^{T}.
For the detection of sequences (i) erroneously annotated as containing BSs, (ii) with shifted BSs, or (iii) with missing or wrong strand annotations, we introduce the three random variables u_{1}, u_{2}, and u_{3}.
The variable u_{1} handles the possibility that a sequence annotated as containing a BS does not contain a BS. u_{1} = 0 denotes the case that the sequence contains no BS, and u_{1} = 1 denotes the case that the sequence contains exactly one BS. If the sequence contains one BS, it can be located at different positions and on both strands.
The variable u_{2} handles the possibility of shifts of a BS caused by annotation errors. u_{2} models the start position of the BS in the sequence with respect to the annotated start position. This variable can assume the integer values {s, (s1), ..., s1, s}, where s is the maximal shift of the BS upstream or downstream of the annotated position.
The variable u_{3} handles the possibility that a BS can have two orientations in the doublestranded upstream region of the target gene. According to the notation of CoryneRegNet, u_{3} = 0 denotes the forward strand defined as the strand in 5'â†’3' direction relative to the target gene, and u_{3} = 1 denotes the reverse complementary strand.
For shortness of notation, we define u := (u_{1}, u_{2}, u_{3}). Because we do not know the values of u, these variables are modeled as hidden variables. We assume that u_{2} and u_{3} are conditionally independent of each other given u_{1}; that is, we assume that annotation errors of position and strand are conditionally independent given the occurrence of the BS. We define
where the subscript h stands for hidden, and where É¸ := (É¸_{1}, É¸_{2}, É¸_{3}) denotes the vector of parameters of this distribution. MotifAdjuster allows the user to specify the probability P_{ h }(u_{1}É¸_{1}) that a sequence contains (or does not contain) a BS and the probability distribution P_{ h }(u_{2}u_{1}, É¸_{2}) for the length of the erroneous shift. In addition, MotifAdjuster estimates the logarithm of the probability that the BS is located on the forward (v = 0) or the reverse complementary (v = 1) strand, , from the userprovided data as described in subsection Expectation maximization algorithm.
The hidden values of u lead to the likelihood
of the data x given the model parameters (Î», Ï„, É¸), where the sum runs over all possible values of u. Here, the subscript a stands for accumulated, and the subscript c stands for composite. In the following, we define the likelihood in close analogy to [8, 37]. If sequence x contains no BS, we assume that x is generated by a homogeneous Markov model of order 0; that is,
If the sequence x contains a BS, then u_{2} encodes its start position, u_{3} encodes its strand, and we assume that the nucleotides upstream and downstream of the BS are generated by a homogeneous Markov model of order 0, yielding
and
where the subscript m stands for motif.
Prior
As prior of the parameters of the PWM model, we use the "common choice" [34â€“36] of a product of transformed Dirichlets
where denotes the positive hyperparameter of , denotes the equivalent sample size (ESS) at position â„“, which we set to be equal at each position, Î±^{â„“}denotes the fourdimensional vector , and Î± denotes the (4 Ã— w) matrix (Î±_{1}, ..., Î±_{ w }).
The choice of this prior is pragmatic rather than biologically motivated. This prior is conjugate to the likelihood, allowing to write the posterior as a product of transformed Dirichlets. As PWM models are special cases of Bayesian networks, the chosen prior can be understood as a special case of the Bayesian Dirichlet (BD) prior [38].
Analogously, for homogeneous Markov models of order 0, we choose a transformed Dirichlet P(Ï„Î²) := D(Ï„Î²), where Î²_{ a }denotes the positive hyperparameter of Ï„_{ a }.
MotifAdjuster allows the user to specify P(u_{1}É¸_{1}) and P(u_{2}u_{1}, É¸_{2}). In principle, MotifAdjuster allows the user to specify any probability distribution P(u_{2}u_{1}, É¸_{2}) for the length of the erroneous shift, allowing also asymmetric or bimodal distributions, if needed. For an easy and userfriendly execution, MotifAdjuster also offers a discrete and symmetrically truncated Gaussian distribution defined by
where z is an integer value ranging from s to s. The realvalued parameter Ïƒ is similar to the standard deviation of a Gaussian distribution and can be specified by the user, and we denote É¸_{2} := (s, Ïƒ).
We expect that some sequences are annotated to contain a BS, although they do not contain a BS in reality, but we believe that the fraction of such incorrectly annotated sequences is small. Hence, we choose P(u_{1} = 0É¸_{1})=0.2 for the studies presented in this article; that is, we assume that only 20% of the sequences annotated to contain a BS do not contain a BS in reality. We further expect that the annotated position of the BS might be shifted accidentally by a few base pairs, so we choose s = 5 and a discrete and symmetrically truncated Gaussian distribution with Ïƒ = 1. This choice results in a conditional probability of approximately 40% that the BS is not shifted, of approximately 25% that it is shifted 1 bp, and of approximately 5% that it is shifted by more than 1 bp upstream or downstream of the annotated start position, respectively, given that a BS is present in sequence x.
As prior of the parameter É¸_{3}, we choose a transformed Dirichlet P(É¸_{3}Î³) := D(É¸_{3}Î³) with Î³ = (Î³_{0}, Î³_{1}), where Î³_{ v }denotes the positive hyperparameter of É¸_{3,v}with v âˆˆ {0, 1}.
Putting all pieces together, we define the prior of the parameters of the mixture model of Equation (4) by:
stating that we assume Î», Ï„, and É¸_{3} to be statistically independent.
We denote the ESS of the mixture model chosen before inspecting any database by Îµ, and we set the ESS of the PWM model to P(u_{1} = 1É¸_{1})Â·Îµ, the positive hyperparameters of the strand parameters to , and the ESS of the homogeneous Markov model of order 0 to (L  P(u_{1} = 1É¸_{1})Â·w)Â·Îµ. For the reassessment of BSs presented in this article, we choose an ESS of Îµ = 5, yielding an ESS of 4 for the PWM model, Î³_{0} = Î³_{1} = 2, and an ESS of 57 for the homogeneous Markov model of order 0. This choice yields for every a âˆˆ {A, C, G, T} and every â„“ âˆˆ [1, w], stating that the chosen prior of the PWM model can be understood as a special case of the BDeu prior [39, 40], which in turn is a special case of the BD prior.
Expectation maximization algorithm
The model parameters of the mixture model defined by Equation (4) cannot be estimated analytically, but any numeric optimization algorithm can be used for maximizing the posterior. One popular optimization algorithm for maximizing the likelihood P(SÎ», Ï„, É¸) is the EM algorithm [41]. The EM algorithm can be easily modified for maximizing the posterior P(Î», Ï„, É¸S, Î±, Î², Î³) of the data set S by iteratively maximizing:
with
Q(Î», Ï„, É¸, Î»^{(t)}, Ï„^{(t)}, É¸^{(t)}Î±, Î², Î³) can be maximized analytically with respect to Î», Ï„, and É¸_{3}, yielding the familiar expressions provided in Additional data file 2. The posterior P(Î», Ï„, É¸S, Î±, Î², Î³) increases monotonically with each iteration, implying that the modified EM algorithm converges to the global maximum, a local maximum, or a saddle point. We stop the algorithm if the logarithmic increase of the posterior between two subsequent iterations becomes smaller than 10^{6}, restart the algorithm 10 times with randomly chosen initial values of , and choose the parameters of that start with the highest posterior, similar to [8, 37]. If we restrict P_{ h }(u_{2}u_{1}, É¸_{2}) to a uniform distribution over all possible start positions, if we set P_{ h }(u_{3}u_{1} = 1) = 0.5, and if we restrict the background model to be strand symmetric, then we obtain the probabilistic model that is the basis of [8, 37].
The flexibility allowed by MotifAdjuster is important for its practical applicability. Typically, the user has prior knowledge about (i) the expected motif occurrence and (ii) the shift distribution, but (iii) no or only limited prior knowledge about the distribution of the BS strand orientation. Hence, we allow the user to specify the logarithm of the probability that a sequence contains a BS É¸_{1,0}, a nonuniform distribution to incorporate the prior knowledge of the shift distribution, and we estimate the logarithm of the probability that the BS is located on the forward strand É¸_{3,0} from the data. This setting allows MotifAdjuster to work, without additional intervention, also in the two extreme cases that the BSs lie predominantly either on the forward or on the reverse complementary strand.
Because of the open source license of MotifAdjuster, similar mixture models can be derived and implemented easily, for instance, by using other background and motif models such as Markov models of higher order [42â€“44], Permuted Markov models [45], Bayesian networks [46, 47], or their extensions to variable order [48â€“53].
Case studies
In this section we present the results of MotifAdjuster applied to seven data sets of Escherichia coli, the validation of MotifAdjuster results for NarL BSs, and the prediction of a novel NarL BS.
Results for seven data sets of Escherichia coli
For testing the efficacy of MotifAdjuster and improving the annotation of BSs of Escherichia coli, we extract all data sets with at least 30 BSs of length of at most 25 bp from the bacterial generegulatory reference database CoryneRegNet 4.0. The choice of at least 30 BSs of length of at most 25 bp is arbitrary, but motivated by the intention that the results of the following study should not be influenced by TFs with an insufficient number of BSs or by TFs with an atypical BS length. Seven data sets of BSs corresponding to the TFs CpxR, Crp, Fis, Fnr, Fur, Lrp, and NarL satisfy these requirements, and we apply MotifAdjuster to each of these seven data sets. We summarize the results obtained by MotifAdjuster in Table 1, and we provide a complete list of the results in Additional data file 3.
We find that all of the data sets are considered questionable by MotifAdjuster and, more surprisingly, that 34.5% of the 536 BS annotations are proposed for removal or shifts. The percentage of questionably annotated BSs ranges from 9.3% for Fnr to 95.7% for Fur. MotifAdjuster proposes to remove 51 of the 536 BSs and to shift 134 of the remaining 485 BSs by at least one bp, indicating that, in these seven data sets, erroneous shifts of the annotated BSs are the most frequent annotation error. In particular, the percentage of proposed deletions ranges from 2.2% (one of 46) for Fur to 27.3% (nine of 33) for CpxR, and the percentage of proposed shifts ranges from 5.6% (three of 54) for Fnr to 93.5% (43 of 46) for Fur. In more detail, we observe a broad range of shift lengths ranging from one shift 4 bp upstream to two shifts 4 bp downstream, with a sharp peak about 0.
For each of the seven TFs, we analyze whether the adjustments proposed by MotifAdjuster result in an improved motif of the BSs (Figure 1). We compute the sequence logos [54, 55] of the original BSs obtained from CoryneRegNet and those of the BSs proposed by MotifAdjuster, which we call original sequence logos and adjusted sequence logos, respectively. Comparing these sequence logos, we find that the adjusted sequence logos show a higher conservation than the original sequence logos in all seven cases. We also compare the sequence logos with consensus sequences obtained from the literature [56â€“61], and we find that the adjusted sequence logos are more similar to the consensus sequences than the original sequence logos. In addition, we find, for the TFs CpxR, Fur, and NarL, that the adjusted sequence logos allow us to recognize clear motifs that could not be recognized in the original sequence logos obtained from CoryneRegNet.
We investigate whether there exists any systematic dependence of the observed rate of proposed adjustments exists on the number of BSs, the BS length, and the GC content of the BSs. We find no obvious dependence of the error rate on the number of BSs and on the BS length. Comparing the GC content of the BSs, we find that the GC content of the BSs of all but one TF ranges from 30% to 40%. However, the GC content of the Fur BSs is only 20%. This low GC content might be the reason for the unexpectedly high percentage of shifts in this data set, because it is more likely to shift a BS accidentally in a sequence composed of a virtually binary alphabet.
Validation of MotifAdjuster results for NarL
To evaluate the previous results, we choose NarL as example and scrutinize the proposed reannotations of MotifAdjuster for this case. The nitrate regulator NarL of Escherichia coli is one of the key factors controlling the upregulation of the nitrate respiratory pathway and the downregulation of other respiratory chains. In the absence of oxygen, the energetically most efficient anaerobic respiratory chain uses nitrate and nitrite as electron acceptors [62]. Detection of and adaptation to extracellular nitrate levels are accomplished by complex interactions of a double twocomponent regulatory system, which consists of the homologous sensory proteins NarQ and NarX, and the homologous TFs NarL and NarP. Depending on the BS arrangement and localization relative to the transcription start site, NarL and NarP act as activators or repressors, thereby enabling a flexible control of the expression of nearly 100 genes.
CoryneRegNet stores 74 NarL BSs, each of length 7 bp (Table 1). Of these 74 BSs, only 36 are considered accurate by MotifAdjuster, whereas 38 are considered to be questionable. In 25 cases, MotifAdjuster proposes to switch the strand orientation of the BS; in five cases, it proposes to shift the location of the BS, and for six BSs, it proposes both a switch of strand orientation and a shift of position. In addition, two BSs are proposed for removal. We present a summary of these results in Table 2, we provide a complete list of the results in Additional data file 4, and we summarize in Table 3 those 13 BSs of the regulator NarL where MotifAdjuster proposes to shift the location of the BS or to remove it from the databases.
To evaluate the accuracy of MotifAdjuster, we check the original literature [63, 37] for each of the 13 questionable BS candidates. Comparing both, we find that the proposed annotations agree with those in the literature in all cases but one (BS of gene b1224). That is, in 12 of 13 cases signaled by MotifAdjuster as being questionable, the detected error was indeed caused by an inaccurate transfer from the original literature into the generegulatory databases RegulonDB and CoryneRegNet. Of those 12 questionable BSs, 10 BSs are correctly proposed to be shifted, and two are correctly proposed to be removed.
Turning to the BS of the gene b1224, we find it is published as given in the databases [64], in contrast to the proposal of MotifAdjuster. However, Darwin et al. [67] report that a mutation of this BS has little or no effect on the expression of b1224. Hence, the proposal could possibly be correct, and the BS could be shifted or even be deleted.
In addition, MotifAdjuster checks the strand annotation of BSs and proposes strand switches if needed. To validate these annotations, we cannot use the annotations from RegulonDB and CoryneRegNet, because these databases contain all BSs in 5'â†’3' direction relative to the target gene. Hence, we consult annotation experts at the Center for Biotechnology in Bielefeld to reannotate the strand orientation of the BSs manually, and we compare the results with those of MotifAdjuster. Interestingly, we find that the strand orientations proposed by MotifAdjuster are in perfect (100%) agreement with the manuallycurated strand orientations. As an independent test of the efficacy of MotifAdjuster for NarL BSs, we use the manually annotated BSs provided by the PRODORIC database [68]. Remarkably, we find also in this case that the results of MotifAdjuster perfectly agree with the annotations.
Another hint that the proposed adjustments of MotifAdjuster could be reasonable is based on the observation that NarL and NarP homodimers bind to a 727' BS arrangement [61], an inverted repeat structure consisting of a BS on the forward strand, a 2bp spacer, and a BS on the reverse complementary strand. NarP exclusively binds as homodimer to this 727' structure. NarL homodimers bind at 727' sites with highaffinity, but NarL monomers can also bind to a variety of other heptamer arrangements. Instances of this 727' structure have been reported for four genes: fdnG, napF, nirB, and nrfA [61, 65]. In contrast to this observation, all BSs in CoryneRegNet as well as RegulonDB are annotated to be on the forward strand, including the second half of the inverted repeat. When applied to these four genes, MotifAdjuster proposes all heptamers of the second half of the 727' structure to be switched to the reverse strand, in agreement with [61, 65]. In addition, MotifAdjuster proposes six additional 727' BS arrangements, located in the upstream regions of the genes adhE, aspA, dcuS, frdA, hcp, and norV. The positions and the orientations are presented in Additional data file 4.
Prediction of a novel NarL binding site
After investigating to which degree MotifAdjuster is capable of finding errors in existing generegulatory databases, it is interesting to test whether MotifAdjuster could be helpful for finding novel BSs. The flexibility of BS arrangements and the low motif conservation complicate the computational and manual prediction of NarL BSs by curation teams. This results in several cases in which promoter regions are experimentally verified to be bound by NarL, but in which no NarL BS could be detected [69, 70]. Examples of such genes are caiF [71], torC [72], nikA [73], ubiC [74], and fdhF [75]. We extract the upstream regions of these genes, where an upstream sequence is defined by CoryneRegNet as the sequence between positions 560 bp and +20 bp relative to the first position of the annotated start codon of the first gene of the target operon. In addition, we extract those upstream regions of Escherichia coli that belong to operons not annotated as being regulated by NarL (background data set).
We investigate whether we can now detect NarL BSs based on the adjusted data set that could not be detected based on the original data set from CoryneRegNet. For that purpose, we estimate the parameters Î» of the PWM model on the adjusted data set as proposed by MotifAdjuster and Ï„ of the homogeneous Markov model on the background data set. From the adjusted PWM, we build a mixture model over both strands with the same probability for each strand; that is, exp(É¸_{3,0}) = exp(É¸_{3,1}) = 0.5. For the classification of an unknown heptamer x, we build a simple likelihoodratio classifier with these parameters Î», Ï„, É¸_{3} and define the loglikelihood ratio by
For an upstream region, we compute r_{ max }defined as the highest loglikelihood ratio of any heptamer x in this upstream region. We compute the P value of a potential BS x with value r(x) as fraction of the background sequences whose r_{ max }values exceed r(x).
With this classifier, a significant NarL BS can now be detected in the upstream region of torC. Figure 2a shows the doublestranded DNA fragment with the predicted BS (TACCCCT) located on the forward strand starting at 209 bp relative to the start codon, and at 181 bp relative to the annotated transcription start site [76]. The distance of the predicted BS to the start codon agrees with the distance distribution of previously known NarL BS (Figure 2b), providing additional evidence for the predicted BS. This finding closes the gap between sequenceanalysis and geneexpression studies, as the torCAD operon consists of three genes that are essential for the trimethylamine Noxide (TMAO) respiratory pathway [76]. TMAO is present as an osmoprotector in tissues of invertebrates and can be used as respiratory electron acceptor by Escherichia coli. Transcriptional regulation of this operon by NarL binding to the proposed BS would explain nitratedependent repression of TMAOterminal reductase (TorA) activity under anaerobic conditions [72], thereby linking TMAO and nitrate respiration.
Conclusions
Generegulatory databases, such as AGRIS, AthaMap, CoryneRegNet, CTCFBSDB, JASPAR, ORegAnno, PRODORIC, RegulonDB, SCPD, TRANSFAC, TRED, or TRRD store valuable information about generegulatory networks, including TFs and their BSs. These BSs are usually manually extracted from the original literature and subsequently stored in databases. The whole pipeline of wetlab BS identification and annotation, publication, and manual transfer from the scientific literature to data repositories is not just time consuming but also error prone, leading to many false annotations currently present in databases.
MotifAdjuster is a software tool that supports the (re)annotation process of BSs in silico. It can be applied as a qualityassurance tool for monitoring putative errors in existing BS repositories and for assisting with a manual strand annotation. MotifAdjuster maximizes the posterior of the parameters of a simple mixture model by considering the possibilities that (i) a sequence being annotated as containing a BS in reality does not contain a BS; (ii) the annotated BS is erroneously shifted by a few base pairs; and (iii) the annotated BS is erroneously located on the false strand and must be reverse complemented. In contrast to existing denovo motifdiscovery algorithms, MotifAdjuster allows the user to specify the probability of finding a BS in a sequence and to specify a nonuniform shift distribution.
We apply MotifAdjuster to seven data sets of BSs for the TFs CpxR, Crp, Fis, Fnr, Fur, Lrp, and NarL with a total of 536 BSs, and we find 51 BSs proposed for removal and 134 BSs proposed for shifts. In total, this results in 34.5% of the BSs being proposed for adjustments. We choose NarL as an example to scrutinize the proposed reannotations of MotifAdjuster. Checking the original literature for each of the 13 cases shows that the proposed deletions and shifts of MotifAdjuster are in agreement with the published data. Comparing the strand annotation of MotifAdjuster with independent information indicates that the proposals of MotifAdjuster are in accordance with human expertise. Furthermore, MotifAdjuster enables the detection of a novel BS responsible for the regulation of the torCAD operon, finally augmenting experimental evidence of its NarL regulation. MotifAdjuster is an opensource software tool that can be downloaded, extended easily if needed, and used for computational reassessments of BS annotations.
Availability and requirements
Project name: MotifAdjuster, project home page: [77], operating system(s): platform independent. Programming language: Java 1.5. Requirements: Jstacs 1.2.2. License: GNU General Public License version 3.
Additional data files
The following additional data are available with the online version of this article. Additional data file 1 contains a comparison of denovo motifdiscovery tools including MEME, RecursiveSampler, Improbizer, SeSiMCMC, AGLAM, and MotifAdjuster for the reannotation of NarL. Additional data file 2 contains a detailed description of the MAP parameter estimators of the model. Additional data file 3 contains a list of MotifAdjuster results for all seven data sets. Additional data file 4 contains a list of MotifAdjuster results compared with the original input of CoryneRegNet and RegulonDB for the TF NarL.
Abbreviations
 BS:

binding site
 EM:

expectation maximization
 ESS:

equivalent sample size
 MAP:

maximum a posteriori
 PWM:

position weight matrix
 TF:

transcription factor.
References
Babu MM, Teichmann SA: Evolution of transcription factors and the gene regulatory network in Escherichia coli. Nucleic Acids Res. 2003, 31: 12341244. 10.1093/nar/gkg210.
Pabo CO, Sauer RT: Transcription factors: structural families and principles of DNA recognition. Annu Rev Biochem. 1992, 61: 10531095. 10.1146/annurev.bi.61.070192.005201.
Hellman LM, Fried MG: Electrophoretic mobility shift assay (EMSA) for detecting proteinnucleic acid interactions. Nat Protoc. 2007, 2: 18491861. 10.1038/nprot.2007.249.
Galas DJ, Schmitz A: DNAse footprinting: a simple method for the detection of proteinDNA binding specificity. Nucleic Acids Res. 1978, 5: 31573170. 10.1093/nar/5.9.3157.
Benotmane AM, Hoylaerts MF, Collen D, Belayew A: Nonisotopic quantitative analysis of proteinDNA interactions at equilibrium. Analyt Biochem. 1997, 250: 181185. 10.1006/abio.1997.2231.
MÃ¶nke G, Altschmied L, Tewes A, Reidt W, Mock HP, BÃ¤umlein H, Conrad U: Seedspecific transcription factors ABI3 and FUS3: molecular interaction with DNA. Planta. 2004, 219: 158166. 10.1007/s0042500412069.
Sun LV, Chen L, Greil F, Negre N, Li TR, Cavalli G, Zhao H, Steensel BV, White KP: ProteinDNA interaction mapping using genomic tiling path microarrays in Drosophila. Proc Natl Acad Sci USA. 2003, 100: 94289433. 10.1073/pnas.1533393100.
Bailey TL, Elkan C: Fitting a mixture model by expectation maximization to discover motifs in biopolymers. Proc Int Conf Intell Syst Mol Biol. 1994, 2: 2836.
Lawrence CE, Altschul SF, Boguski MS, Liu JS, Neuwald AF, Wootton JC: Detecting subtle sequence signals: a Gibbs sampling strategy for multiple alignment. Science. 1993, 262: 208214. 10.1126/science.8211139.
Thompson W, Rouchka EC, Lawrence CE: Gibbs Recursive Sampler: finding transcription factor binding sites. Nucleic Acids Res. 2003, 31: 35803585. 10.1093/nar/gkg608.
Ao W, Gaudet J, Kent WJ, Muttumu S, Mango SE: Environmentally induced foregut remodeling by PHA4/FoxA and DAF12/NHR. Science. 2004, 305: 17431746. 10.1126/science.1102216.
Favorov AV, Gelfand MS, Gerasimova AV, Ravcheev DA, Mironov AA, Makeev VJ: A Gibbs sampler for identification of symmetrically structured, spaced DNA motifs with improved estimation of the signal length. Bioinformatics. 2005, 21: 22402245. 10.1093/bioinformatics/bti336.
Kim NK, Tharakaraman K, MarinoRamirez L, Spouge J: Finding sequence motifs with Bayesian models incorporating positional information: an application to transcription factor binding sites. BMC Bioinformatics. 2008, 9: 26210.1186/147121059262.
Baumbach J, Wittkop T, Kleindt CK, Tauch A: Integrated analysis and reconstruction of microbial transcriptional gene regulatory networks using CoryneRegNet. Nature Protocols. 2009,
MÃ¼nch R, Hiller K, Barg H, Heldt D, Linz S, Wingender E, Jahn D: PRODORIC: prokaryotic database of gene regulation. Nucleic Acids Res. 2003, 31: 266269. 10.1093/nar/gkg037.
GamaCastro S, JimÃ©nezJacinto V, PeraltaGil M, SantosZavaleta A, PeÃ±alozaSpinola MI, ContrerasMoreira B, SeguraSalazar J, MuÃ±izRascado L, MartÃnezFlores I, Salgado H, BonavidesMartÃnez C, AbreuGoodger C, RodrÃguezPenagos C, MirandaRÃos J, Morett E, Merino E, Huerta AM, TreviÃ±oQuintanilla L, ColladoVides J: RegulonDB (version 6.0): gene regulation model of Escherichia coli K12 beyond transcription, active (experimental) annotated promoters and Textpresso navigation. Nucleic Acids Res. 2008, 36: D120D124. 10.1093/nar/gkm994.
Palaniswamy SK, James S, Sun H, Lamb RS, Davuluri RV, Grotewold E: AGRIS and AtRegNet: a platform to link cisregulatory elements and transcription factors into regulatory networks. Plant Physiol. 2006, 140: 818829. 10.1104/pp.105.072280.
BÃ¼low L, Engelmann S, Schindler M, Hehl R: AthaMap, integrating transcriptional and posttranscriptional data. Nculeic Acids Res. 2009, 37: D983D986. 10.1093/nar/gkn709.
Bao L, Zhou M, Cui Y: CTCFBSDB: a CTCFbinding site database for characterization of vertebrate genomic insulators. Nucleic Acids Res. 2008, 36: D83D87. 10.1093/nar/gkm875.
Sandelin A, Alkema W, EngstrÃ¶m P, Wasserman WW, Lenhard B: JASPAR: an openaccess database for eukaryotic transcription factor binding profiles. Nucleic Acids Res. 2004, 32: D91D94. 10.1093/nar/gkh012.
Montgomery SB, Griffith OL, Sleumer MC, Bergman CM, Bilenky M, Pleasance ED, Prychyna Y, Zhang X, Jones SJM: ORegAnno: an open access database and curation system for literaturederived promoters, transcription factor binding sites and regulatory variation. Bioinformatics. 2006, 22: 637640. 10.1093/bioinformatics/btk027.
Zhu J, Zhang M: SCPD: a promoter database of the yeast Saccharomyces cerevisiae. Bioinformatics. 1999, 15: 607611. 10.1093/bioinformatics/15.7.607.
Matys V, KelMargoulis OV, Fricke E, Liebich I, Land S, BarreDirrie A, Reuter I, Chekmenev D, Krull M, Hornischer K, Voss N, Stegmaier P, LewickiPotapov B, Saxel H, Kel AE, Wingender E: TRANSFAC and its module TRANSCompel: transcriptional gene regulation in eukaryotes. Nucleic Acids Res. 2006, 34: D108D110. 10.1093/nar/gkj143.
Jiang C, Xuan Z, Zhao F, Zhang MQ: TRED: a transcriptional regulatory element database, new entries and other development. Nucleic Acids Res. 2007, 35: D137D140. 10.1093/nar/gkl1041.
Kolchanov NA, Ignatieva EV, Ananko EA, Podkolodnaya OA, Stepanenko IL, Merkulova TI, Pozdnyakov MA, Podkolodny NL, Naumochkin AN, Romashchenko AG: Transcription Regulatory Regions Database (TRRD): its status in 2002. Nucleic Acids Res. 2002, 30: 312317. 10.1093/nar/30.1.312.
Kel AE, GÃ¶ssling E, Reuter I, Cheremushkin E, KelMargoulis OV, Wingender E: MATCH: a tool for searching transcription factor binding sites in DNA sequences. Nucleic Acids Res. 2003, 31: 35763579. 10.1093/nar/gkg585.
Tompa M, Li N, Bailey TL, Church GM, Moor BD, Eskin E, Favorov AV, Frith MC, Fu Y, Kent WJ, Makeev VJ, Mironov AA, Noble WS, Pavesi G, Pesole G, RÃ©gnier M, Simonis N, Sinha S, Thijs G, van Helden J, Vandenbogaert M, Weng Z, Workman C, Ye C, Zhu Z: Assessing computational tools for the discovery of transcription factor binding sites. Nat Biotechnol. 2005, 23: 137144. 10.1038/nbt1053.
Beckstette M, Homann R, Giegerich R, Kurtz S: Fast index based algorithms and software for matching position specific scoring matrices. BMC Bioinformatics. 2006, 7: 38910.1186/147121057389.
MÃ¼nch R, Hiller K, Grote A, Scheer M, Klein J, Schobert M, Jahn D: Virtual Footprint and PRODORIC: an integrative framework for regulon prediction in prokaryotes. Bioinformatics. 2005, 21: 41874189. 10.1093/bioinformatics/bti635.
Stormo G, Schneider T, Gold L, Ehrenfeucht A: Use of the "Perceptron" algorithm to distinguish translational initiation sites. Nucleic Acids Res. 1982, 10: 29973010. 10.1093/nar/10.9.2997.
Staden R: Computer methods to locate signals in nucleic acid sequences. Nucleic Acids Res. 1984, 12: 505519. 10.1093/nar/12.1Part2.505.
Bernardo JM, Smith AFM: Bayesian Theory. 1994, New York: John Wiley & Sons
Thiesson B: Accelerated quantification of Bayesian networks with incomplete data. Proceedings of First International Conference on Knowledge Discovery and Data Mining (KDD95): August 2021 1995. Edited by: Fayyad U, Uthurusamy R. 1995, Montreal: AAAI Press, 306311.
MacKay DJ: Choice of basis for Laplace approximation. Machine Learning. 1998, 33: 7786. 10.1023/A:1007558615313.
Heckerman D: A Tutorial on Learning with Bayesian Networks. Tech. Rep. MSRTR9506, Microsoft Research. 1995
Meila M, Jordan MI: Learning with mixtures of trees. J Machine Learning Res. 2000, 1: 148. 10.1162/153244301753344605.
Lawrence CE, Reilly AA: An expectation maximization (EM) algorithm for the identification and characterization of common sites in unaligned biopolymer sequences. Proteins Struct Funct Genet. 1990, 7: 4151. 10.1002/prot.340070105.
Cooper GF, Herskovits E: A Bayesian method for the induction of probabilistic networks from data. Machine Learning. 1992, 9: 309347.
Buntine WL: Operations for learning with graphical models. J Artific Intelligence Res. 1994, 2: 159225.
Heckerman D, Geiger D, Chickering D: Learning Bayesian Networks: The Combination of Knowledge and Statistical Data. 1995, Tech. rep., Microsoft Research, Redmond, WA: Advanced Technology Division
Dempster AP, Laird NM, Rubin DB: Maximum likelihood from incomplete data via the EM algorithm. J R Stat Soc Series B. 1977, 39: 122.
Zhang M, Marr T: A weight array method for splicing signals analysis. Comput Appl Biosci. 1993, 9: 499509.
Salzberg SL: On comparing classifiers: pitfalls to avoid and a recommended approach. Data Mining Knowledge Discov. 1997, 1: 317328. 10.1023/A:1009752403260.
Thijs G, Lescot M, Marchal K, Rombauts S, De Moor B, Rouze P, Moreau Y: A higherorder background model improves the detection of promoter regulatory elements by Gibbs sampling. Bioinformatics. 2001, 17: 11131122. 10.1093/bioinformatics/17.12.1113.
Ellrott K, Yang C, Sladek FM, Jiang T: Identifying transcription factor binding sites through Markov chain optimization. Bioinformatics. 2002, 18: S100S109. 10.1093/bioinformatics/18.1.100.
Barash Y, Elidan G, Friedman N, Kaplan T: Modeling dependencies in proteinDNA binding sites. Proceedings of Seventh Annual International Conference on Computational Molecular Biology. 2003, 2837.
Castelo R, Guigo R: Splice site identification by idlBNs. Bioinformatics. 2004, 20: i6976. 10.1093/bioinformatics/bth932.
Rissanen J: A universal data compression system. IEEE Trans Inform Theory. 1983, 29: 656664. 10.1109/TIT.1983.1056741.
Ron D, Singer Y, Tishby N: The power of amnesia: learning probabilistic automata with variable memory length. Machine Learning. 1996, 25: 117149. 10.1023/A:1026490906255.
Boutilier C, Friedman N, Goldszmidt M, Koller D: Contextspecific Independence in Bayesian networks. Proceedings of the Twelfth Annual Conference on Uncertainty in Artificial Intelligence. 1996, 115123.
BÃ¼hlmann P: Model Selection for Variable Length Markov Chains and Tuning the Context Algorithm. 1997, Tech. Rep. 82, Statistics, Zurich: ETH Zentrum
Zhao X, Huang H, Speed TP: Finding short DNA motifs using permuted Markov models. J Comput Biol. 2005, 12: 894906. 10.1089/cmb.2005.12.894.
BenGal I, Shani A, Gohr A, Grau J, Arviv S, Shmilovici A, Posch S, Grosse I: Identification of transcription factor binding sites with variableorder Bayesian networks. Bioinformatics. 2005, 21: 26572666. 10.1093/bioinformatics/bti410.
Schneider TD, Stephens RM: Sequence logos: a new way to display consensus sequences. Nucleic Acids Res. 1990, 18: 60976100. 10.1093/nar/18.20.6097.
Crooks GE, Hon G, Chandonia JM, Brenner SE: WebLogo: a sequence logo generator. Genome Res. 2004, 14: 11881190. 10.1101/gr.849004.
De Wulf P, McGuire AM, Liu X, Lin ECC: Genomewide profiling of promoter recognition by the twocomponent response regulator CpxRP in Escherichia coli. J Biol Chem. 2002, 277: 2665226661. 10.1074/jbc.M203487200.
KÃ¶rner H, Sofia HJ, Zumft WG: Phylogeny of the bacterial superfamily of CrpFnr transcription regulators: exploiting the metabolic spectrum by controlling alternative gene programs. FEMS Microbiol Rev. 2003, 27: 559592. 10.1016/S01686445(03)000664.
Pan CQ, Johnson RC, Sigman DS: Identification of new Fis binding sites by DNA scission with Fis1,10phenanthrolinecopper(I) chimeras. Biochemistry. 1996, 35: 43264333. 10.1021/bi952040z.
Baichoo N, Helmann JD: Recognition of DNA by Fur: a reinterpretation of the Fur box consensus sequence. J Bacteriol. 2002, 184: 58265832. 10.1128/JB.184.21.58265832.2002.
Cui Y, Wang Q, Stormo G, Calvo J: A consensus sequence for binding of Lrp to DNA. J Bacteriol. 1995, 177: 48724880.
Maris AE, KaczorGrzeskowiak M, Ma Z, Kopka ML, Gunsalus RP, Dickerson RE: Primary and secondary modes of DNA recognition by the NarL twocomponent response regulator. Biochemistry. 2005, 44: 1453814552. 10.1021/bi050734u.
Unden G, Bongaerts J: Alternative respiratory pathways of Escherichia coli: energetics and transcriptional regulation in response to electron acceptors. Biochim Biophys Acta. 1997, 1320: 217234. 10.1016/S00052728(97)000340.
Kaiser M, Sawers G: Nitrate repression of the Escherichia coli pfl operon is mediated by the dual sensors NarQ and NarX and the dual regulators NarL and NarP. J Bacteriol. 1995, 177: 36473655.
Li J, Kustu S, Stewart V: In vitro interaction of nitrateresponsive regulatory protein NarL with DNA target sequences in the fdnG, narG, narK and frdA operon control regions of Escherichia coli K12. J Mol Biol. 1994, 241: 150165. 10.1006/jmbi.1994.1485.
Darwin AJ, Tyson KL, Busby SJ, Stewart V: Differential regulation by the homologous response regulators NarL and NarP of Escherichia coli K12 depends on DNA binding site arrangement. Mol Microbiol. 1997, 25: 583595. 10.1046/j.13652958.1997.4971855.x.
Golby P, Kelly DJ, Guest JR, Andrews SC: Transcriptional regulation and organization of the dcuA and dcuB genes, encoding homologous anaerobic C4dicarboxylate transporters in Escherichia coli. J Bacteriol. 1998, 180: 65866596.
Darwin AJ, Li J, Stewart V: Analysis of nitrate regulatory protein NarLbinding sites in the fdnG and narG operon control regions of Escherichia coli K12. Mol Microbiol. 1996, 20: 621632. 10.1046/j.13652958.1996.5491074.x.
PRODORIC URL of the Matrix of NarL. [http://www.prodoric.de/matrix.php?matrix_acc=MX000003]
Overton TW, Griffiths L, Patel MD, Hobman JL, Penn CW, Cole JA, Constantinidou C: Microarray analysis of gene regulation by oxygen, nitrate, nitrite, FNR, NarL and NarP during anaerobic growth of Escherichia coli: new insights into microbial physiology. Biochem Soc Trans. 2006, 34: 104107. 10.1042/BST0340104.
Constantinidou C, Hobman JL, Griffiths L, Patel MD, Penn CW, Cole JA, Overton TW: A reassessment of the FNR regulon and transcriptomic analysis of the effects of nitrate, nitrite, NarXL, and NarQP as Escherichia coli K12 adapts from aerobic to anaerobic growth. J Biol Chem. 2006, 281: 48024815. 10.1074/jbc.M512312200.
Eichler K, Buchet A, Lemke R, Kleber HP, MandrandBerthelot MA: Identification and characterization of the caiF gene encoding a potential transcriptional activator of carnitine metabolism in Escherichia coli. J Bacteriol. 1996, 178: 12481257.
Iuchi S, Lin EC: The narL gene product activates the nitrate reductase operon and represses the fumarate reductase and trimethylamine Noxide reductase operons in Escherichia coli. Proc Natl Acad Sci USA. 1987, 84: 39013905. 10.1073/pnas.84.11.3901.
Rowe JL, Starnes GL, Chivers PT: Complex transcriptional control links NikABCDEdependent nickel transport with hydrogenase expression in Escherichia coli. J Bacteriol. 2005, 187: 63176323. 10.1128/JB.187.18.63176323.2005.
Kwon O, DruceHoffman M, Meganathan R: Regulation of the ubiquinone (coenzyme Q) biosynthetic genes ubiCA in Escherichia coli. Curr Microbiol. 2005, 50: 180189. 10.1007/s0028400444171.
Wang H, Gunsalus RP: Coordinate regulation of the Escherichia coli formate dehydrogenase fdnGHI and fdhF genes in response to nitrate, nitrite, and formate: roles for NarL and NarP. J Bacteriol. 2003, 185: 50765085. 10.1128/JB.185.17.50765085.2003.
MÃ©jean V, IobbiNivol C, Lepelletier M, Giordano G, Chippaux M, Pascal MC: TMAO anaerobic respiration in Escherichia coli: involvement of the tor operon. Mol Microbiol. 1994, 11: 11691179. 10.1111/j.13652958.1994.tb00393.x.
Jstacs: A Java Framework for Statistical Analysis and Classification of Biological Sequences. [http://www.jstacs.de]
Acknowledgements
We thank Lothar Altschmied, Helmut BÃ¤umlein, Karina Brinkrolf, Linda GÃ¶tz, Jan Grau, Astrid Junker, Gudrun MÃ¶nke, Michaela Mohr, Stefan Posch, Yvonne PÃ¶schl, Sven Rahmann, Michael Seifert, Marc Strickert, and Andreas Tauch for helpful discussions, two anonymous reviewers for their valuable comments, Alexander Goesmann, Achim Neumann, and Ralf Nolte for expert technical support, and Richard MÃ¼nch for his help with the RegulonDB data. J.B. greatly appreciates the support of the German Academic Exchange Service (DAAD). This work was supported by grant 0312706A by the German Ministry of Education and Research (BMBF) and XP3624HP/0606T by the Ministry of Culture of SaxonyAnhalt.
Author information
Authors and Affiliations
Corresponding author
Additional information
Authors' contributions
JK and IG developed the basic idea, and JK implemented MotifAdjuster. JB and TK provided the data. All authors contributed to data analysis, writing, and approved the final manuscript.
Electronic supplementary material
13059_2009_2196_MOESM1_ESM.pdf
Additional data file 1: Comparison of denovo motifdiscovery tools including MEME, RecursiveSampler, Improbizer, SeSiMCMC, AGLAM, and MotifAdjuster for the reannotation of NarL. (PDF 32 KB)
13059_2009_2196_MOESM4_ESM.txt
Additional data file 4: List of MotifAdjuster results for the TF NarL compared with the original input of CoryneRegNet and RegulonDB. (TXT 3 KB)
Authorsâ€™ original submitted files for images
Below are the links to the authorsâ€™ original submitted files for images.
Rights and permissions
About this article
Cite this article
Keilwagen, J., Baumbach, J., Kohl, T.A. et al. MotifAdjuster: a tool for computational reassessment of transcription factor binding site annotations. Genome Biol 10, R46 (2009). https://doi.org/10.1186/gb2009105r46
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1186/gb2009105r46
Keywords
 Additional Data File
 Expectation Maximization Algorithm
 TMAO
 Sequence Logo
 Position Weight Matrix