The Flavor network

The Flavor Network represents the graph of flavor sharing across all Entities/Natural Ingredients, which are shown along the circumference. To observe the pattern of sharing, click on any Entity. As the flavor network is way too dense, To address dense pattern of interrelationships, the backbone network with statistically significant edges is depicted; further only entities sharing more than four molecules are shown, entities having 15 or less edges were exempted from this condition. The entities are grouped together according to categories (separated by spaces), making it visually easier to observe flavor sharing within and across categories.

To access ‘The Flavor Network’, please click on the ‘Load the Flavor Network’ button on the main page.

It is interesting to observe unexpected similarities among seemingly disparate entities, which can be explored further using the Search options.

FlavorDB Visual Search

Visual Search, as the name suggests, provides a graphic way of exploring FlavorDB. Starting with a map that depicts all the categories, one could click on any of the boxes to open up a map of ingredients in that category.

For example clicking on the category ‘Dairy’ will show a detailed map of all the ingredients in it. Further hovering over at ingredient ‘Butter’, for example, will pop up a tooltip (small box) with a link to the page with entity details. Clicking on the URL will open show the list of all 131 molecules in Butter, along with details of the entity as well as its natural source.

At any stage, to go back one level back, just ‘right click’ anywhere on the map.

Flavor Molecules Search

As part of main search tool, ‘Flavor Molecules’ search facilitates querying and browsing FlavorDB2 on the basis of a host of features: Common Name, FEMA No., CAS No., Functional Group, FEMA Flavor Profile, Flavor Profile, Description of the Molecule, Regulatory Status (IOFI, CoE, Trade Association Guidelines), Natural Occurrence, Aroma/Taste Threshold Values, Synthesised by, Food Category, Molecular Weight, Number of Hydrogen Bond Donor/Acceptors, Type of Molecules (Natural, Synthetic, Source Unknown). Apart from that one could use JSME Molecular Editor for generating a query by drawing a structure of a molecule. Other than hand-drawing the structure, one could also paste SMILES or upload Mol or SDF files.

This search allows conjunction of all the query parameters provided. Therefore the query could be refined by using multiple constraints. The search generates the Results page listing all flavor molecules in the database that fulfil the criteria.

In case you are not looking for anything specific, just hit the ‘Search’ button without providing any specific query. This will yield in a result showing all the 25595 molecules in FlavorDB. Similarly by choosing ‘Type of Molecules’ to be ‘Natural’, you will get a list of 2254 molecules that are reported to be found in specific natural entities. All search boxes come with an autocomplete option to assist you with any query containing a string with two or more characters typed, greatly enabling the user’s ability to explore the database.

In case you would like to filter the database based on a more nuanced list of constraints, use the ‘Go To Advanced Search’ button.

Flavor Molecules Search : Results

In response to the query parameter provided, a list of all matching flavor molecules will be shown. The Results page displays the query used on the top of the page. In case multiple fields are used for creating a query, the results will show molecules that satisfy all conditions. The list displays following information: Common Name, PubChem ID, FEMA Flavor Profile, Flavor Profile and ‘More Info’ button. Clicking on this button will open the ‘Molecular & Flavor Profile’ information panel for the corresponding molecules.

NOTE: In case you use the JSME tool to search based on the molecular structure, an additional column titled ‘Similarity (0-1)’ is also displayed. This refers to the percentage similarity of the result molecules to the query structure drawn. The minimum similarity threshold for displaying molecules is set at 30%.

Flavor Molecules Search : Synthesis Results

The query performed in the image is for 'Synthesised by'. This field consists the synthesis information of the molecules. It searches for both the molecules and the process a molecule is made from. This field uses '@' for an AND operation and '!' for an OR operation. So for example, the query put in is 'aldehyde@condensation'.

In response to the query parameter provided, a list of all matching flavor molecules will be shown which are synthesised by condensation and require aldehyde for their synthesis.

Flavor Molecules Search : Food Category Results

The query performed in the image is for 'Food Category'. Food Category refers to the category of the food in which the molecule is used. Just like 'Synthesised by', this field uses '@' for an AND operation and '!' for an OR operation. So for example, the query put in is 'Fish products!Cheese'.

In response to the query parameter provided, a list of all matching flavor molecules will be shown which are used in Fish products or Cheese or both.

Flavor Molecules Search : Natural Occurrence Results

The query performed in the image is for 'Natural Occurrence'. Natural Occurrence signifies the natural entities in which the molecule is found. For example, the query put in is grapes. Just like 'Synthesised by', this field uses '@' for an AND operation and '!' for an OR operation.

In response to the query parameter provided, a list of all matching flavor molecules will be shown which are present in grapes.

Molecular Information Panel

This panel provides ‘Molecular & Flavor Profile’ of the molecule. This includes molecular identity such as Common Name, IUPAC Name, SMILES, Empirical Formula, Molecular Weight, CAS ID, PubChem ID, FEMA Flavor Profile, FEMA Number, FLAVOR PROFILE. As part of Flavor Profile, it provides information about Taste, Odor and Functional Groups.

Further, the following collapsed sections are available, which provide following details on click: Physicochemical Properties, ADMET Properties, Nomenclature, 2D/3D Properties, Description, Consumption, Regulatory Status, Specifications, Aroma/Taste Threshold Values, Reported Uses, Natural Occurrence, Synthesis, Composition, Physical-Chemical Characteristics of the molecule, and list of all entities that contain the flavor molecule.

The panel also provides 2D image of the molecule, using JSmol applet, which transforms into 3D view upon clicking ‘View JSmol’ button. Further, an option is provided to download the molecule data in Mol2, SDF, 2D image, JSON format. Importantly, using structural similarity search one could either explore similar molecules within FlavorDB, or could identify commercially available molecules from ZINC.

Advanced Search

‘Advanced Search’ is provided to enable users who would like to probe FlavorDB2 by viewing the data through refined parameters. The modality of ‘Advanced Search’ is similar to that of ‘Molecular/Main Search’ which allows filtering FlavorDB2 data by conjunction of parameters, including graphic query by drawing a query chemical structure.

In addition to the broad parameters, it facilitates use of queries based on refined molecular features such as NAS No., FL No., CoE No., EINECS No., JECFA No., Number of Aromatic Rings, Energy etc.

Entity Search

Apart from exploring the flavor space through molecular parameters, FlavorDB facilitates search through the Entity/Ingredient. ‘Entity’ refers to a natural ingredient belonging any of the broad categories (such as Spice, Dairy, Fruit, Plant, Vegetable, Bakery, Flower etc.). Entity search could be formed based on ‘Entity Name’ as well as ‘Category Name’.

Importantly, we facilitate search based on synonyms knowing very well that many ingredients are known by wide variety of names. For example, to search for ‘Eggplant’, you may also search by either ‘Aubergine’ or ‘Brinjal’. Looking for ‘Custard Apple’, is as good as looking for ‘Cherimoya’. The search box comes with an autocomplete option to assist you with any ingredient containing a string matching with two or more characters typed, greatly enabling the user’s ability to navigate the database.

Natural Source Search

FlavorDB extends the information of flavor molecules beyond their Entity/Ingredient, to their natural sources. These include details of common name of the source with more technical details such as Kingdom, Family, Genus and Species of the source. Again, autocomplete assists in easing out search as well as wandering around.

Entity Details Page

In response to ‘Entity Search’ as well as ‘Natural Source Search’ the Entity Details page returns the Entity page. This page provides information about the Entity as well as its Natural Source. For the Entity, details of its category, its picture, synonyms and link to its Wikipage are provided; whereas for its Natural Source, its name, image, link to Wikipage, and details of scientific classification extracted from Wikipage are provided. The Entity page contains list of flavor molecules along with ‘Molecular & Flavor Profile’ of each molecule.

Flavor Pairing Search

User can pair an entity by entering the entity name in the given field.

Importantly, we facilitate search based on synonyms knowing very well that many ingredients are known by wide variety of names. For example, to search for ‘Eggplant’, you may also search by either ‘Aubergine’ or ‘Brinjal’. Looking for ‘Custard Apple’, is as good as looking for ‘Cherimoya’. The search box comes with an autocomplete option to assist you with any ingredient containing a string matching with two or more characters typed, greatly enabling the user’s ability to traverse the database.

If you wish to pair two ingredients in the database, you can first search for one of the two ingredients you wish to pair. For example, if you wish pair parsnip and tomato, search for either parsnip or tomato. The complete flavor pairing analysis for search entity will be displayed along with ingredients with which it shares flavor molecules. The user can then select or search for the ingredient in ‘detailed flavor pairing analysis’ table.

Flavor Pairing

‘Flavor Pairing’, also known as ‘Food Pairing’, is a heuristic used for finding ingredient pairs that are expected to go well together. Historically such pairing has been practiced largely based on trial and error, and relies heavily on human judgment and intuition of connoisseurs. Evidence-based understanding of rules or heuristics that dictate choice of food facilitates better informed experiments to test hypothesis, and hopefully to design acceptable food pairings.

Founded on the extensive repertoire of 937 entities and their constituent flavor molecules (2254), ‘Food Pairing’ tool provides a powerful engine for spanning through more than 437000 pairings to reveal interesting flavor profile overlaps across categories and disparate entities. The Flavor Network provides a bird’s eye view of this information.

When one searches for a specific ingredient in ‘Flavor Pairing’ app, the results are presented in a hierarchy. At the basic level, it provides information of ingredient and its natural source, along scientific classification of the latter. It also provides graphic results for flavor pairing at the level of categories as well as ingredients.

For example, Parsnip (a root vegetable) has most sharing with categories, Fruit, Spice, Dairy etc. At the ingredient level, it shows highest sharing with Tomato, Ginger, Carrot, Beer etc.

The ‘Detailed Flavor Pairing Analysis’ section gives complete list of ingredients that share one or more flavor molecules with the query Entity/Ingredient. Clicking on ‘Entity Name’ or corresponding ‘Number of Shared Flavor Molecules’ opens up details of shared molecules. One may as well probe the results through search box provided with the results panel.

Receptors

In the absence of comprehensive information in literature that link flavor molecules with their receptors, we provide a list of 33 receptors known to be involved in gustatory sensation. These include receptors for Sweet (3), Bitter (25), Sour (3), and Umami (2). We also provide a list of 1068 odor receptors.

The panel includes details of ‘Receptor Name’, ‘Uniprot ID’, and link to Uniprot page. The outgoing link takes one to comprehensive information of the receptor, including sequence, structure, function, taxonomy, pathology, expression, interaction, and list of similar proteins. These details are useful in exploring the interaction of flavor molecules with receptors as well as probing the mechanism of action responsible for flavor sensation.

Molecule of the Day

‘Molecule of the Day’ feature provides peek into FlavorDB by randomly picking one flavor molecule every day. The user is invited to examine the 3D structure of the molecules (View JSmol), to go through its ‘Molecular & Flavor Profile’, and to search the FlavorDB space for structurally similar molecules.

FlavorDB Statistics: Molecules

‘FlavorDB Statistics’ are provided on the landing page of FlavorDB. The ‘Molecules’ tab provides an overview of database, by way of distribution of categories of entities, ‘Number of Unique Molecules’ in the category and ‘Number of Entities’ in it.

Each blob represents a category and is color coded to depict ‘Average Number of Molecules per Ingredient’. Clicking on any blob would pop up details of the Category indicating statistics of ‘Number of unique molecules in the category’, ‘Number of ingredients in the category’ and ‘Average number of molecules per ingredient'.

FlavorDB Statistics: Entities

The Pie Chart shows number of entities in each of the categories. Hovering over the chart pops up name of the entity category as well as number of entities in it.

FlavorDB Statistics: Flavors

This chart provides frequency of Flavor terms for all flavor compounds reported to be found in natural ingredients. For clarity of visualization, only flavor terms that occur ten or more times are shown. Hovering over data points pops up the name of the Flavor term and its frequency. This can be used as a sneak peek into the flavor characterization of the database, to further search the database using any of the desired flavor term.

FlavorDB Statistics: Functional Groups

This chart provides frequency of Functional Group across FlavorDB. Hovering over data points pops up the name of the Functional Group and its frequency. Starting with an overview of the functional characterization of molecules from this chart, one could then explore the database by using one of the groups as a search query.