LeTICE [85] was also used as an alternative method for comparison with our approach. LeTICE is not a generic clustering method but is designed specifically for the problem of genetic regulatory network prediction. It is based on integrating TF binding data with expression pattern data to define a genetic regulatory network, i.e. a set of clusters each comprising genes with a common TF binding pattern and a shared pattern of expression. This is achieved by finding the network, 𝐵, which maximizes
𝑃 (𝐵|𝐿, 𝐸)
where 𝐿 is a matrix of TF binding probabilities and 𝐸 a matrix of gene expression patterns. LeTICE location matrix consists of rows corresponding to genes and columns corresponding to TFs. According to LeTICE, the p-value (for the hypothesis that there is no interaction between a TF and the promoter of a gene) will be more pragmatic than deciding on a threshold for p-values.
As such LeTICE is a method based on a similar premise of integrating TF binding data and expression data to find regulatory relationships, but being based on different underlying methodology it is an ideal comparator, albeit only relevant to the problem of genetic regulation. To provide a direct comparison of algorithms, LeTICE was applied to the dataset described above. Note that LeTICE takes binding p values directly as input and that it has its own TF and gene pre-selection criteria, in this case it selected 18 TFs and 289 genes. LeTICE was then run with the optimum runtime parameters suggested in the original paper.
As part of this study we also examined the effect of using normalized (where each gene was normalized to zero mean and unit standard deviation) and un-normalized gene expression data. We also compared joint clustering to clustering expression data separately, which can be done by simply omitting binary variables in the input to our program.
Table 3.1
The filtered TFs and their functional information.With the exception of TFs in blue, the rest of the TFs are with known involvement in cell cycle. The functional descriptions were retrieved from the
Saccharomyces cerevisiae Genome Database (SGD) [96].
TF Cell cycle related functions
ACE2 Sequence-specific DNA binding RNA polymerase II transcription factor involved in G1/S transition of the mitotic cell cycle; activates cytokinetic cell separation; also regulates antisense transcription at diverse loci; localizes to both nucleus and cytosol
ASH1 Sequence-specific DNA binding RNA Pol II transcription factor that up and down regulates transcription; its role as transcription repressor negatively regulates mating type switching, G1/S transition of mitotic cell cycle; its role as transcription activator positively regulates pseudo- hyphal growth; subunit of the Rpd3L histone deacetylase complex; also localizes to the cellular bud
DAL80 A RNA polymerase II transcription factor that binds specific DNA sequence; negatively regulates transcription and is involved in nitrogen catabolite repression of transcription; localized to the nucleus
FKH1 Sequence-specific DNA binding transcription factor involved in chromatin remodeling, mitotic transcription regulation, transcription termination, mating-type switching, and pseudo hyphal growth; binds DNA replication origins and positively regulates replication initiation; also binds centromeres
FKH2 RNA polymerase II transcription factor involved in positive and negative regulation of transcription during mitotic cell cycle; positively regulates DNA replication initiation; binds replication origins and promoters in sequence-specific manner; localizes to cytosol and nucleus
GZF3 GATA zinc finger protein; negatively regulates nitrogen catabolic gene expression by competing with Gat1p for GATA site binding; function requires a repressive carbon source; dimerizes with Dal80p and binds to Tor1p
Table 3.1
The filtered TFs and their functional information.With the exception of TFs in blue, the rest of the TFs are with known involvement in cell cycle. The functional descriptions were retrieved from the
Saccharomyces cerevisiae Genome Database (SGD) [96]. (Continued)
TF Cell cycle related functions
MBP1 Sequence-specific DNA binding transcription factor that positively regulates transcription by RNA polymerase II involved in the G1/S transition of mitosis; subunit of the MBF (Mlu1 cell cycle box Binding Factor) transcription complex
MCM1 Transcription factor; involved in cell-type-specific transcription and pheromone response; plays a central role in the formation of both repressor and activator complexes; relocalizes to the cytosol in response to hypoxia
MET31 Zinc-finger DNA-binding transcription factor; targets strong transcriptional activator Met4p to promoters of sulfur metabolic genes; involved in transcriptional regulation of the methionine biosynthetic genes
NDD1 Transcriptional activator essential for nuclear division; localized to the nucleus; essential component of the mechanism that activates the expression of a set of late-S-phase-specific genes; turnover is tightly regulated during cell cycle and in response to DNA damage
PDR1 Sequence specific DNA-binding polymerase II transcription factor that activates expression of genes involved in drug response
RCS1 Sequence-specific DNA binding transcription factor that regulates chromatid cohesion, chromosome segregation, and cellular iron homeostasis; localizes to the cytoplasm, nucleus, and kinetochores
STB1 Protein with role in regulation of MBF-specific transcription at Start; phosphorylated by Cln-Cdc28p kinases in vitro; un-phosphorylated form binds Swi6p, which is required for Stb1p function; expression is cell-cycle regulated
SWI4 DNA binding component of the SBF complex (Swi4p-Swi6p); a transcriptional activator that in concert with MBF (Mbp1-Swi6p) regulates late G1-specific transcription of targets including cyclins and genes required for DNA synthesis and repair
Table 3.1
The filtered TFs and their functional information.With the exception of TFs in blue, the rest of the TFs are with known involvement in cell cycle. The functional descriptions were retrieved from the
Saccharomyces cerevisiae Genome Database (SGD) [96]. (Continued)
TF Cell cycle related functions
SWI5 Transcription factor that recruits Mediator and Swi/Snf complexes; activates transcription of genes expressed at the M/G1 phase boundary and in G1 phase; required for expression of the HO gene controlling mating type switching; localization to nucleus occurs during G1 and appears to be regulated by phosphorylation by Cdc28p kinase
SWI6 Transcription cofactor; forms complexes with Swi4p and Mbp1p to regulate transcription at the G1/S transition; in volved in meiotic gene expression; also binds Stb1p to regulate transcription at START
YHP1 Homeobox transcriptional repressor; binds Mcm1p and early cell cycle box (ECB) elements of cell cycle regulated genes, thereby restricting