TRED was originally designed as a resource for studies on gene function and regulation. It provides cis-elements, such as promoters and binding motifs, and trans-elements, such as transcriptional factors (TFs). The promoters of target genes came from two sources: experimental determination and sequence-based computational prediction. These two sources complement each other. In TRED, hand curation was applied as a crucial part of the data collection to ensure data accuracy. Based on the reliability of the supporting evidence for each promoter, a quality level was assigned.

One key feature of TRED is the easy access to interaction data between TFs and the promoters of their target genes, including binding motifs reported by the previous studies. Although part of the data was obtained from the existing gene regulation resources, most of the data came from our exhaustive literature curation. The TF-binding motifs were mapped on to the promoter sequences of their target genes. Along with the binding motifs, the experimental evidence and other important pertinent information were also collected. A binding quality level was assigned based on definitiveness of the binding evidence, which was determined by the experimental approaches employed to demonstrate the binding and data interpretation from experts. For example, we assigned ‘known’ as the binding quality level to a binding that has been proved by gel-shift competition, DNase I footprinting, etc.

In order to provide users with more complete information of the genes, cross-references to other well-known database such as PubMed, GenBank, GeneCards (1) and TRANSFAC (2) were established in TRED. A comprehensive description of the content and the structure of TRED has been published earlier (3). In addition, many on-the-fly tools were implemented for the analysis of sequences retrieved from TRED as well as imported from other resources. The user interface and software functionality were also described in the previous report (3).

Table 1 Number of promoters and genes in TRED, with gene numbers in parentheses

The human genome codes for ~1850 TFs, which account for 6.0% of its estimated total number of genes (8). It is a daunting task to collect and curate comprehensive and precise interaction data between the TFs and their target genes. Since cancer is one of the greatest threats to human health and has been a field under extensive study, including a broad interest in understanding cell cycle regulatory networks, we started out by focusing on target genes of cancer-related TFs. Previously, TRED contained mainly the target genes and promoters for TF families E2F and Myc (3). In this new release, we expanded it to 34 new TF families that have been implicated in cancer pathways, including p53, AP1, ER and NFκB/Rel. They are involved in many cellular processes, such as proliferation, differentiation, cell motility and apoptosis. There are totally 9308 newly collected target genes (5365 in human, 2526 in mouse and 1417 in rat) and 10251 target promoters (5956 in human, 2736 in mouse and 1559 in rat) for these TF families. The detailed distribution of the target promoters and the target genes is listed in Table 2.

Table 2 Number of curated target promoters/genes for the 36 TF families

Although TRANSFAC also provides factor-site interaction data, it contains less information for the TFs and their target genes available in TRED. Its latest version (version 7.0) has collected 1040 factors and 608 genes for human and 765 factors and 417 genes for mouse (2). Therefore, on average each factor has less than one target gene. In contrast, the number of target genes per TF in TRED is much higher. For example, there are >200 target genes on average for each TF family for human in TRED (Table 2). This can provide fairly resolved gene regulatory networks (GRNs) for the 36 TF families involved in cancer pathways. In addition to this, TRED contains relatively complete genome-wide promoter annotation for human, mouse and rat. Moreover, the binding sites in TRED were also mapped on to the assembled chromosomes. These absolute genomic positions make it ready to associate TRED with other genomic data for various studies. However, it should be noted that TRANSFAC also collects factor–site interaction data of species other than human, mouse and rat. Therefore, although TRED and TRANSFAC overlap to certain extent, they complement each other at some aspects.

Figure 1 Sample pages showing access to the gene regulatory network of TF family GLI (glioma-associated oncogene homolog) in human. Ellipses, TFs; and squares, genes. Arrows indicate interactions between two genes. Red arrows imply that the binding quality level (more ...)

The website (http://rulai.cshl.edu/TRED) offers the following services: easy access to TRED entries through text-based query interface; search for the target genes of a given TF; retrieval of the promoter sequences and the TF-binding motifs; further analysis of the retrieved sequences of promoters and motifs. TRED homepage also provides the access to the GRNs of the TF families in human, mouse and rat, which were constructed from its collection of the interaction data between the TFs and their target genes.

We thank Dr Thomas Kubarych for the critical reading of the manuscript for English. This work is supported by a contract grant from NCI and partly by a NIH grant HG001696. Funding to pay the Open Access publication charges for this article was provided by NIH/NHGRI.

Conflict of interest statement. None declared.