The authors have declared that no competing interests exist.
Conceived and designed the experiments: TH JAR. Performed the experiments: TH BK KL IT MS NK RS TR AM. Analyzed the data: TH JAR JP JK AM. Contributed reagents/materials/analysis tools: JP JK. Wrote the paper: TH JAR.
Computational modeling of biological processes is a promising tool in biomedical research. While a large part of its potential lies in the ability to integrate it with laboratory research, modeling currently generally requires a high degree of training in mathematics and/or computer science. To help address this issue, we have developed a web-based tool, Bio-Logic Builder, that enables laboratory scientists to define mathematical representations (based on a discrete formalism) of biological regulatory mechanisms in a modular and non-technical fashion. As part of the user interface, generalized “bio-logic” modules have been defined to provide users with the building blocks for many biological processes. To build/modify computational models, experimentalists provide purely qualitative information about a particular regulatory mechanisms as is generally found in the laboratory. The Bio-Logic Builder subsequently converts the provided information into a mathematical representation described with Boolean expressions/rules. We used this tool to build a number of dynamical models, including a 130-protein large-scale model of signal transduction with over 800 interactions, influenza A replication cycle with 127 species and 200+ interactions, and mammalian and budding yeast cell cycles. We also show that any and all qualitative regulatory mechanisms can be built using this tool.
With the goal of understanding the complexities of various biological processes, computational modeling is an important part of Systems Biology. However, despite the excitement around computational systems biology and its potential, it has been difficult to fully utilize modeling as part of laboratory research. This is largely due to a significant gap between the computational and experimental sides of the science
In this paper, we present a new tool, Bio-Logic Builder, which allows those without technical knowledge in modeling to build and modify complex computational, qualitative models without the need to write or edit any mathematical equations. Becuase models created in Bio-Logic Builder utilize a commonly used logical (Boolean) mathematical framework (e.g.,
The presented Bio-Logic Builder was successfully tested on one of the largest computational models of signal transduction
Biological interactions defined using the Bio-Logic builder are described by Boolean expressions that users build by using qualitative descriptives (or “bio-logic” components) generally used by laboratory scientists to explain the interaction from experimental studies. Leveraging the qualitative nature in which many biochemical interactions are discovered, Bio-Logic Builder provides users with building blocks of two types. First, users can define modules corresponding to positive and/or negative regulators that are involved in a given biological interaction (e.g., kinase X phosphorylates and activates protein Y, as is the case in studies of biochemical signal transduction). Because only few biological interactions can be represented as simple positive and/or negative regulators, users can specify a second type of bio-logic modules. These modules – “conditions” and “subconditions” – allow users to describe regulatory mechanism in which the effects of one or more positive and/or negative regulators depend on an additional regulators step (e.g., localization, priming, co-factors etc.), and hence the activation state or presence (or absence) of an additional regulator (or group of regulators). As a result, the users can define complex positive and negative regulatory modules much in the same way biological data and knowledge are discovered in the laboratory. To demonstrate how Bio-Logic Builder is used to build biological regulatory mechanisms, in this section is presented a case study which centers around the construction of a relatively complex regulatory system of the signaling protein Rac. Note that a simpler example of how the tool can be used can be viewed in a tutorial video at
Rac is an important player in the regulation of many cellular processes such as cell migration, cytoskeletal reorganization, DNA synthesis, etc. Rac belongs to the Rho family of small guanosine triphosphatases (GTPases), a subgroup of the Ras superfamily. Rac becomes activated when bound to GTP, a process mediated by guanine nucleotide exchange factors (GEFs). The hydrolysis of GTP to GDP results in the inactive state of Rac. This conversion occurs via Rac's intrinsic GTPase activity and is further accelerated by GTPase-activating proteins (GAPs). However, in addition to GAPs and GEFs, Rac's activity also depends on its proper localization as well as the activity state of components of other signaling pathways. A summary of the intricacies involved in the (de-)activation mechanism of Rac as reported in the biochemical literature so far follows. (Note that the following regulatory mechanism of Rac reflects the optimized mechanism published as part of a validated large-scale model of signal transduction in a generic fibroblast cell
In the aforementioned fibroblast model, Rac is defined as ON when it is GTP-bound and localized in the plasma membrane. (See
As one can see, the regulatory mechanism of Rac is intricate and involves a large number of upstream regulators. Specifically, the activation states of 13 upstream regulators, in addition to the activation state of Rac in the previous time point have to be considered, resulting in 14 regulating inputs of Rac. Thus the truth table representation of the function would require the scientist to manually fill out
The Bio-Logic Builder tool is part of The Cell Collective modeling suite
As the name suggests, Negative Regulation Center is where users designate upstream regulation modules that have a negative (i.e., inhibitory) effect on the species of interest (Rac in this example). From the regulatory mechanism described above, the negative regulators of Rac include Akt, RalBP1, p190RhoGAP, and RhoGDI. As shown in
A) Main page of the Negative Regulation Center. B) RalBP1 condition page. In order to define the condition RalBP1 is a negative regulator of Rac only when Rac is on, the user first selects the IF/THEN clause. In order to specify Rac as the conditioned species, the user can drag it from the Species Palette into the indicated gray box in the main part of the screen. Finally, the conditioned state of Rac (“is ON”) needs to be selected. In addition, as indicated by the green buttons, the user can subsequently i) save the condition and either return to the Negative Regulation Center page or add another condition for RalBP1, ii) discard this condition, or iii) add one or more sub-conditions that will be attached to the specified condition of RalBP1. The red Previous button takes the user to the previous screen, whereas the Discard All Conditions & Go Back button removes all conditions and returns the user to the Negative Regulation Center. C) Negative Regulation Center of Rac with all modules fully defined.
Once Akt, RalBP1, p190RhoGAP, and RhoGDI are designated as negative regulation modules, conditions can be specified. As discussed in the previous sections, conditions allow biologists to specify regulatory scenarios under which a particular upstream regulator is dependent on the activity state of another species (e.g., a co-factor). In our Rac example, RalBP1 and p190RhoGAP are responsible for removing GTP from a GTP-bound (i.e., active) Rac and replacing it with GDP, hence inactivating Rac. Therefore, the effects of these negative regulators are dependent on the activation state of Rac itself which can be represented as a condition for the two upstream regulators. Based on the context of the whole network in
The conditions page is accessed from the Negative (Positive) Regulation Center page by clicking on the Center of the negative (positive) regulation module for which conditions need to be added/modified. For example, the conditions page for RalBP1 can be accessed under the “RalBP1 Center”. Users can define conditions as IF/WHEN and UNLESS statements to define scenarios when the effects of the regulator for which a condition is being specified depend on the activity state of another biological species. As mentioned above, the condition associated with Akt, RalBP1, and p190RhoGAP is that these species are negative regulators IF/WHEN Rac is ON. In addition, for users' convenience, each condition can be annotated to reflect its biological meaning and context. In this case, the condition was named “Rac activity”, but any annotation can be used. In the case of RhoGDI, its effect on the activity of Rac depends on the presence/absence of PAK; specifically, RhoGDI acts as a negative regulator UNLESS PAK is ON. Note that any number of conditions (and subconditions) can be associated with any regulator, allowing for the definition of the most complex regulatory mechanisms. (An example of multi-condition scenario is presented in the next section.) See
Positive regulation modules of the species of interest (e.g., Rac) are specified in the Positive Regulation Center. As discussed at the beginning of the Case Study section, the activating species of Rac include RasGRF, Tiam, Pix/Cool, and DOCK180. Once these species have been added to the Species Palette, they can be defined as positive regulation modules in a similar fashion as was done with the negative regulation modules, and was demonstrated in the Negative Regulation Center section. As the regulatory mechanism suggests, all positive regulators are dependent on cell attachment, and hence the activity of ECM and Integrins. Therefore the positive regulation modules RasGRF, Tiam, and DOCK180 the condition (named Cell attachment) “IF/WHEN ECM AND Integrins are ON”. However, the conditions associated with the positive module Pix/Cool are more complicated. As discussed at the beginning of this case study, there are three nontrivial scenarios describing the role of Pix/Cool in the regulation of Rac activity. To capture this complex regulatory mechanism, both condition and subcondition bio-logic gates are necessary. Subconditions can be specified after clicking the “Subconditions” button on the condition page of the regulation module center.
The three scenarios differ based on the presence and absence of G
As discussed in the main text, the regulation of Rac by Pix/Cool is associated with three different scenarios described by three conditions (and subconditions which are not displayed). When multiple conditioned species are added as part of a condition (as is the case with the first condition of the Pix/Cool module, where the activation states of both PAK and G
Once all negative and positive regulation modules are defined, the user is led to the next screen, the Dominance Page. On this page, users can define the “strength” of each negative module in terms of how dominant it is over the individual positive regulation modules. A negative module dominant over all positive regulators (pre-selected by default) has the largest (negative) effect on the state of the species of interest, whereas a negative module dominant over none of the positive modules will have no effect on the activity of the species.
Once the strength of the negative regulation modules is selected, the user needs to specify the final component of the regulatory mechanism building process – the state (active/inactive) of the species in the case where none of the positive nor negative modules are active or present in the cell (model). Upon the last page and component of the Bio-Logic Builder tool, the user can navigate to the Summary Page (
In Bio-Logic Builder, the head regulators represent the main positive/negative regulation modules, within which conditions and subconditions are subsequently added. In the Rac case study presented in this section, the head regulators of the negative regulation modules included Akt, RalBP1, p190RhoGAP, and RhoGDI, whereas the head regulators of the positive regulation modules constituted RasGRF, Tiam, Pix/Cool, and DOCK180 (
While for many biological interactions it is clear (based on the available published data) which of the species is considered the “head regulator”, there are many instances in which regulatory mechanisms can be ambiguous and hence become confusing to the user of Bio-Logic Builder. These few regulatory mechanisms can even be relatively simple in terms of the number of species involved in the interaction. For example, consider a hypothetical biochemical signaling protein X with two phosphorylation sites in its regulatory region. Let's assume that, in order to be fully activated, both of the phosphorylation sites of X need to be phosphorylated, one by kinase Y and the other one by kinase Z. From this described situation, one could consider both kinases as “equal” rather than as a “head regulator”/“condition” relationship. However, based on the Rac case study and the previous discussions of the Bio-Logic Builder algorithm, one of the kinases (Y or Z) has to be considered the head regulator, whereas the other one is represented as a condition (IF/WHEN ON) as part of the regulation module. Which way this scenario should be defined, however, is not clear in this example. Nonetheless, it is important to note that when a number of regulation species appear to be conceptually equal, Bio-Logic Builder requires one of these species be selected as the head regulator whereas the others be considered as a (sub)condition(s). Fortunately, because of the mathematical relationship between the head regulators and the conditions, the mathematical representation of the interaction will be the same in both cases (as detailed in Supporting Information S1). If such a scenario arises, the user will need to use their discretion and decide how to represent the regulatory mechanism.
The
The usability and intuitiveness of Bio-Logic Builder was tested by creating some of the most complex regulatory mechanisms included in one of the largest models of signal transduction (135 molecular species with hundreds of biochemical interactions,
Because models constructed in the presented tool are described using standard Boolean formalisms, they can be simulated by a large number of software tools. In addition to the previously mentioned Cell Collective (whose simulation engine is based on ChemChains
The lack of simple-to-use tools for creating/editing and simulating computational models plays a significant role in the gap that exists between the computational and experimental sides of biomedical research
Bio-Logic Builder is a server-based tool implemented in Java and powered by MySQL database. The user interface was implemented primarily using JavaServer Faces (
Boolean modeling is a (kinetic) parameter-free modeling approach based on qualitative data (e.g., kinase
Which state a node will assume at any given time is determined by the node's Boolean function. Boolean functions can be represented in various ways such as truth tables or Boolean expressions. As an example, consider a hypothetical simple two-node model in
A) Static diagram representing the relationship between the nodes. B) Truth table representations of the Boolean functions for nodes P and Q.
The truth tables depicting regulatory mechanisms of nodes
At the most basic level of a biological regulatory mechanism, users can define positive regulation and negative regulation modules (activator and inhibitors, respectively). Users also define the dominance of individual negative regulators over positive regulators (if applicable). Finally, users specify the state of the biological entity in the absence (i.e., inactivity) of all positive and negative regulators. These regulatory definitions are subsequently used by the software to create the Boolean function representing a regulatory mechanism of a given species as follows.
All main positive/negative regulators are defined as independent modulators of the given species. Hence the Boolean expression is constructed by concatenating all positive regulators using the Boolean OR operator (
Users may define complex regulatory mechanisms using conditions and subconditions that are applied to a (positive/negative) regulation module. Each positive or negative regulator can have
When defining a condition or subcondition, multiple regulators may be specified. In this case, the user must specify the relationship between the regulators. This may be co-operative or independent. As illustrated in Panels B and C in
As mentioned in the
Each (sub)condition is defined by combining the “IF/WHEN” or “UNLESS” statements with a list of regulators that are specified as either Active or Inactive. In terms of a Boolean expression, the specifying IF/WHEN Inactive or UNLESS Active for a given (set of) regulators corresponds to the Boolean NOT operator. (As mentioned above, multiple (sub)condition regulators are defined, the user must specify whether the regulators are co-operative or independent.
For example, the condition “IF/WHEN A, B, C ARE OFF” (A, B, and C are co-operative regulators) corresponds the following Boolean expression:
Once a user defines the regulatory mechanism of a biological species of interest, Bio-Logic Builder generates the corresponding Boolean function in the form of a Boolean expression as well as a truth table which can be saved in text and tab-delimited (.csv) files, respectively. Because both formats are standard representations of Boolean functions, they can be subsequently read and evaluated by other simulation software tools. Furthermore, the regulatory mechanisms defined in Bio-Logic Builder can be also exported in the Systems Biology Markup Language format (SBML
Bio-Logic Builder is freely available as part of The Cell Collective modeling platform (
(PDF)