The Blobulator: A Toolkit for Identification and Visual Exploration of Hydrophobic Modularity in Protein Sequences

TL;DR

The Blobulator is a new toolkit for identifying and visualizing hydrophobic modules in protein sequences, offering web, plugin, and command line tools for analysis.

Contribution

Introduces the Blobulator toolkit for hydrophobic module detection and visualization in proteins.

Findings

Blobulator uses the blobulation algorithm to detect hydrophobic modules in proteins.

The toolkit includes a webtool, VMD plugin, and command line interface for accessibility.

Applications demonstrate its use in globular, membrane, and disordered proteins.

Abstract

While contiguous subsequences of hydrophobic residues
are essential
to protein structure and function, as in the hydrophobic core and
transmembrane regions, there are no current bioinformatics tools for
module identification focused on hydrophobicity. To fill this gap,
we created the blobulator toolkit for detecting,
visualizing, and characterizing hydrophobic modules in protein sequences.
This toolkit uses our previously developed algorithm, blobulation,
which was critical in both interpreting intraprotein contacts in a
series of intrinsically disordered protein simulations (Lohia et al.,
2019) and defining the “local context” around disease-associated
mutations across the human proteome (Lohia et al., 2022). The blobulator toolkit provides accessible, interactive, and
scalable implementations of blobulation. These are available via a
webtool, a visual molecular dynamics (VMD) plugin, and a command line
interface. We highlight use cases for visualization, interaction analysis,
and modular annotation through three example applications: a globular
protein, two orthologous membrane proteins, and an intrinsically disordered
protein. The blobulator webtool can be found at www.blobulator.branniganlab.org, and the source code with pip installable command line tool, as
well as the VMD plugin with installation instructions, can be found
on GitHub at www.GitHub.com/BranniganLab/blobulator.

Figures

Blobulation algorithm. (A) First, the sequence is digitized. Residues are classified as either hydrophobic (blue) or nonhydrophobic (orange) by comparing their hydropathy to the user-selected threshold, H*. (B) The sequence is then segmented into h-blobs, s-blobs, and p-blobs based on H* and L min. Used with permission from ref , Copyright 2022 PNAS.

Screenshots from the blobulator webtool results page for the insulin sequence (UniProt ID: P01308). Parameter and sequence adjustments: (A) hydropathy scale dropdown menu, (B) hydropathy cutoff H* selection, (C) minimum h- or p-blob length adjustment L min selection, (D) custom mutation panel, (E) additional data and display options, (F) track showing smoothed hydropathy per residue. Remaining panels show blobs colored by various calculated blob properties, described in more detail in Text S3: (G) blob type (H) net charge per residue (I) Das–Pappu phase (J) predicted dSNP enrichment or depletion (K) position on the Uversky–Gillepse–Fink boundary plot (L) predicted fraction of disordered residues (M) Colormaps for panels I (left), J (middle, used with permission from ref , Copyright 2022 PNAS), and K (right). Black triangles on all tracks indicate known disease-associated mutations; clicking the triangles introduces the mutation, while hovering provides a link to the entry for the mutation in dbSNP.

Screenshot of the VMD blobulator plugin. The user sets the Mol ID of the protein (A), provides an atomselection for the display (B), selects a hydropathy scale and may select the auto update option to snap the threshold to a scale-dependent default (C), sets the minimum blob length L min (D) and hydropathy cutoff H* (E), and selects how blobs are colored (F). The “Blobulate” button initiates blobulation and creates graphical representations within VMD (G). All representations are deleted by clicking the “Clear representations” button (H). The “Graphical Representations” control panel contains the created graphical representations, which can be further modified (I), and are eventually displayed in the example viewer window (J) showing Leptin (PDB: 1ax8).

Blobulation of lysozyme. (A) Blobs colored according to the blob type, as outputted from the blobulator webtool, and produced using default settings (H* = 0.4, L min = 4). Annotations indicate catalytic residues (red) and the substrate binding site (purple). (B) Molecular image of lysozyme (PDB: 148L) with peptidoglycan (green) colored by the substrate binding site (left, residues found within 7 Å of peptidoglycan) or by the blob type (right, h-blob: blue, p-blob: orange, s-blob: green). (C) Blobulation under increasingly stringent settings (H* = 0.4 and 0.5, L min = 4 and 8). H-blobs are shown as surfaces. Molecular images were generated in VMD , using the VMD plugin introduced in 3.3.

Blob groups in T4 lysozyme. (A) Structural view (PDB: 2LZM) blobulated using default settings (H* = 0.4, L min = 4). Groups are gray; ungrouped h-blobs are blue, and p-blobs are orange ribbons. (B) Blobulation of T4 lysozyme under increasingly stringent settings. Annotations indicate blob identifiers and blob groups. Molecular images were generated in VMD. ,

Globular tendency of T4 lysozyme blobs. Blobulation using default parameters (H* = 0.4, L min = 4). Blobs for each sequence are colored by Das–Pappu phase, as in Figure I. Red diamonds indicate mutated residues. S117 V, which increases thermostability, joins two h-blobs into one. T157I, which decreases thermostability, creates a new h-blob (as well as an s-blob).

Full text

Introduction

1

Protein sequences are modular: they consist of contiguous units (such as alpha helices, functional domains, etc.), which are incorporated into a range of analysis pipelines, visualization tools, and conceptual frameworks. Few tools for detecting modularity, however, explicitly incorporate residue hydrophobicity, despite the cooperative nature of the hydrophobic effect; the critical role of hydrophobicity in stabilizing the core of globular proteins; and the commonplace use of residue hydrophobicity in predictors of protein disorder ?−? ? ? ? ? and membrane interactions. ?,? We introduced “blobulation” as a scheme for segmenting protein sequences by contiguous hydrophobicity, motivated by the need for an analogue to secondary structure when reducing the dimensionality of intrinsically disordered protein (IDP) simulations.? Although blobulation was first applied to IDPs, contiguously hydrophobic regions are most frequently found in the buried cores of structured proteins and have functional and evolutionary signatures in the human proteome: disease-associated single-nucleotide polymorphisms are more likely to be found in contiguously hydrophobic regions and are particularly likely to bridge two such regions.? These results support the interpretation of hydrophobic blobs as evolutionarily constrained interaction “nodes” within a protein sequence. Here, we introduce a toolkit designed to make blobulation accessible and convenient for a wide range of users.

Blobulation detects modularity in protein sequences by searching for “h-blobs”: subsequences longer than a threshold length in which the predicted hydrophobicity score for every residue (its “hydropathy”) exceeds a user-provided threshold. The remaining subsequences are sorted into “p-blobs” (which satisfy the length but not the hydropathy criterion) and “s-blobs” (which satisfy neither length nor hydropathy criteria). While secondary structure detectors typically use a default set of parameters that users rarely change, blobulation settings are meant to be tuned: adjusting the hydropathy and length thresholds in blobulation allows users to gradually shift from detecting many small modules to a few longer ones, bringing the relevant aspects of sequence organization into focus (use-case example shown in section). Furthermore, since blobulation requires only the sequence, it is particularly appropriate in scenarios where circumventing structure is desirable (including IDPs).

Many existing tools cluster residues into protein modules using secondary structure elements. While blobulation yields modules analogous to secondary structure elements, we have not previously provided “blob” versions of tools that traditionally rely on a secondary structure. On the left half of Table, we list the common use-cases for secondary structure identification among currently available tools: residue characterization (secondary structure of each residue), module identification (determining where an α helix, β sheet, or other secondary structure element begins and ends), and module characterization (calculating collective properties of the secondary structure element, for instance, the net charge per residue within a helix or β sheet). The columns show the properties used to define modularity (secondary structure or hydropathy) and then are subdivided by the functionality of the tools: whether the tool makes modularity predictions (usually via a command-line interface) or provides a graphical interface (via a sequence or structure view). For secondary structure tools, module prediction can be accessed through command-line tools, ?,?,? graphically displayed alongside the sequence, ?−? ? ? ? or incorporated into visualizations of the structure. ?−? ? As shown on the right half of Table, tools for characterization,? annotation, ?−? ? ? or coloring by the hydropathy of individual residues ?,?,? are well-established. Yet, few tools incorporate “modular” definitions based on contiguous hydrophobicity, and we are unaware of any that allow graphical exploration or module characterization. Additionally, while tools that provide plots of residue hydropathy may invite the viewer to identify contiguous hydrophobic regions by the eye, they do not provide a systematic, automatable version of the same process.

1: Features of the Blobulator Toolkit Compared to Other Available Tools for Characterizing Protein Elements by Secondary Structure or Hydropathy

To provide a previously unavailable analogue to secondary structure tools that defines and characterizes modules based on hydropathy (bottom right of Table), we developed the blobulator toolkit: an accessible, interactive, and scalable suite of tools that includes a webtool, command line interface (CLI), and a Visual Molecular Dynamics (VMD)? plugin. The webtool supports a sequence-level view of blobs that includes smooth real-time parameter adjustment, a feature to introduce mutations, and layers of annotations, including physicochemical properties and disease-associated mutations. The back-end program for the webtool is available as a pip-installable CLI and supports high-throughput batch blobulation of FASTA files. Finally, the VMD plugin allows users to view blobs on protein structures by creating a blobulator interface in VMD, which stores identified blobs for analysis across simulation trajectories and the generation of publication-quality images and movies. As a whole, the blobulator toolkit provides previously unavailable functionality for prediction and visualization of hydrophobicity-defined modules and their physicochemical properties. Additionally, the toolkit goes beyond secondary structure analogues, enabling end users to access module characterization through graphical user interfaces (GUIs).

In this article, we outline the blobulation algorithm and each component of the blobulator toolkit. We then demonstrate its utility in an example application: detecting blobs corresponding to tertiary interactions in a globular protein. Two additional applications are provided in Text S4. We close by describing future applications for these tools, including novel insights that the blobulator is positioned to enable.

Blobulation

2

Whole-sequence blobulation consists of two steps: digitization and clustering. We provide a description of the algorithm below as well as a pseudocode version of the algorithm in Text S2.

• 1.In the digitization step, the algorithm defines a given residue as hydrophobic or nonhydrophobic as follows:
• aThe user selects a normalized (0 to

1. hydropathy scale, and sets a threshold hydropathy (H*) on this scale.

• bThe hydropathy for each residue (i) is assigned based on the selected normalized hydropathy scale. The hydropathies are then smoothed for each residue i and the residues adjacent to it in sequence (i – 1, i, and i + 1), yielding the smoothed hydropathy H _ i _.
• cResidue i is classified as hydrophobic (H _ i _ > H*) or nonhydrophobic (H _ i _ ≤ H*), shown in FigureA.
• 2.In the clustering step, subsequences termed “blobs” are defined as follows:
• aThe user sets L min equal to a positive integer representing a threshold number of residues required to form an h- or p-blob.
• bSubsequences of at least L min sequential hydrophobic residues are classified as hydrophobic blobs (h-blobs) (FigureB).
• cAll other linking sequences are then classified based on their L min as either nonhydrophobic blobs (p-blobs, L ≥ L min) or short blobs (s-blobs, L < L min) (FigureB).

Blobulation algorithm. (A) First, the sequence is digitized. Residues are classified as either hydrophobic (blue) or nonhydrophobic (orange) by comparing their hydropathy to the user-selected threshold, H. (B) The sequence is then segmented into h-blobs, s-blobs, and p-blobs based on H* and L min. Used with permission from ref , Copyright 2022 PNAS.*

Though there are default settings for the initial blobulation, the user can tune the various parameters. Different combinations of H* and L min will detect blobs with varying properties, and adjusting them can extract blobs that correspond to key hydrophobic regions (some examples are shown in section). Alternatively, users may reveal hierarchical layers of organization by blobulating a given sequence under varying parameter settings. Very high thresholds (H* approaching 1) will return one large p-blob, and very low thresholds (H* approaching 0) will return one large h-blob, while intermediate thresholds will reveal intrinsic segmentation within the protein sequence. Additionally, high L min will yield only h-blobs if used with a low to moderate H*.

We also consider higher-order organization beyond individual blobs: “blob groups” are h-blobs separated only by s-blobs. Examples and further discussion of blob groups can be found in section. Blobs are labeled as follows: by their type (h, p, s), group number (1, 2, 3), and for h-blobs within a blob group, by a subgroup letter (a, b, c). This forms a unique label for every blob (h1a, s1, h1b, etc.). We refer to the distribution of blobs in a protein sequence as its “blob topology”.

The Blobulator Toolkit

3

Here, we present three tools by which an amino acid sequence can be blobulated: a webtool, a CLI, and a VMD plugin. As shown in the schematic in Figure S1, all three tools share the core blobulation algorithm outlined in section. The webtool can be used to interactively explore blobulated sequences, introduce mutations, and investigate various blob properties. The CLI is the backend for the webtool but can also be used to run batch processes, and it accepts DNA sequences. The VMD plugin can be used with molecular structures for analysis and creating images and movies of blobulated proteins in VMD. All images in this paper reflect the output of blobulator version 1.0b.

Webtool

3.1

Blobulation of any amino acid sequence can be achieved using the blobulator webtool found at www.blobulator.branniganlab.org. On the “New Query” tab of the homepage, the user can submit either a database ID or manually enter a protein sequence for blobulation. From here, a user can blobulate any protein or amino acid sequence using an accepted ID or manually enter the sequence directly as well as toggle to other information tabs. Figure illustrates the blobulation of insulin, a peptide hormone that promotes glucose absorption by cells.

Screenshots from the blobulator webtool results page for the insulin sequence (UniProt ID: P01308). Parameter and sequence adjustments: (A) hydropathy scale dropdown menu, (B) hydropathy cutoff H selection, (C) minimum h- or p-blob length adjustment L min selection, (D) custom mutation panel, (E) additional data and display options, (F) track showing smoothed hydropathy per residue. Remaining panels show blobs colored by various calculated blob properties, described in more detail in Text S3: (G) blob type (H) net charge per residue (I) Das–Pappu phase (J) predicted dSNP enrichment or depletion (K) position on the Uversky–Gillepse–Fink boundary plot (L) predicted fraction of disordered residues (M) Colormaps for panels I (left), J (middle, used with permission from ref , Copyright 2022 PNAS), and K (right). Black triangles on all tracks indicate known disease-associated mutations; clicking the triangles introduces the mutation, while hovering provides a link to the entry for the mutation in dbSNP.*

After the initial blobulation, the user can interactively tune some of the parameters. For example, while our previous studies, ?,? the example applications in this paper, and the webtool default all use the Kyte–Doolittle scale,? the user may select the Eisenberg–Weiss scale? or the Moon–Fleming scale.? Additionally, the user can change the hydropathy scale used for digitization (FigureA), or interactively modify the H* threshold and L min cutoff (here we use H* = 0.4 and L min = 4, respectively) to segment the protein into modules with varying hydropathy and length properties. This can be done by adjusting the respective sliders, or manually entering a value (FigureB,C). A user may also introduce a mutation by selecting one of the black triangles or manually entering the position and alternate amino acid in the “mutate residue” field and selecting the checkbox (FigureD). All changes update dynamically. Users can download both an image of the output as a PDF and the raw data (in CSV format) used to generate the webtool output (FigureE).

The smoothed hydropathy of each residue H _ i _ (as defined in section) is shown in the first “Results” track (FigureF) along with the H* threshold (blue line). While a graph displaying the hydropathy of individual residues is common to many tools, blobulation clusters adjacent residues by their hydrophobicity, revealing higher order organization. The subsequent tracks display blobs colored by various biochemical properties, and a more detailed description is provided in the SI.

Python Scripting and Command Line Interface

3.2

For use in high-throughput applications, like those presented in ref ?, we also provide a stand-alone Python package for blobulation. This package is also the backend for the blobulator webtool and is called when a protein is submitted via that interface. The package can be used either in Python scripts or directly via a CLI, which is pip-installable:

The CLI accepts either amino acid (default) or DNA (-DNA) sequences, either as plain text (default) or as a path to a FASTA file (-Fasta). For a plain text sequence:

For a FASTA file:

The outputs of the CLI are CSV files (one for each input sequence; in the above example, this is named my_blobulation.csv) identical to the downloadable blobulation output of the webtool (see 3.1). Additional example scripts are provided on blobulator GitHub.

VMD Plugin

3.3

While the webtool allows users to view the blobulation of the sequence and various averaged physicochemical properties, the plugin tool for VMD? allows users to incorporate blob information into the visualization and analysis of protein structure and dynamics. Users can assign blobs with the same parameters and algorithm in the webtool while taking advantage of the existing VMD tools for scriptable analyses and creation of high-quality images and movies. Additionally, the blobulation algorithm was reimplemented in TCL, the native VMD scripting language, for maximum compatibility. The blobulation.tcl script extracts the sequence information from the provided structure and then detects blobs based on user-provided parameters. The user, user2, and user3 fields associated with each atom store the type of blob (h, p, and s), the blob ID, or the group ID, respectively. This information can then be used in atom selections or user-field coloring schemes provided by the VMD software; the molecular images in this paper were created by creating an atom selection for each h-blob, representing it with “quick-surf”, and coloring each h-blob a slightly different shade of blue. Examples of images created using this plugin tool can be found throughout the applications in this paper (Figures and ?). For proteins with many blobs, this approach can require many separate graphical representations; Blob_GUI.tcl includes a function for creating representations in a batch process.

Since users of VMD typically create representations through GUIs, we also provide a GUI through which the user can provide and adjust parameters as well as call the scripts that blobulate the sequence and create the representations. This GUI is similar to the control panel found on the blobulator webtool’s output page. There is an option to blobulate only part of a protein using the Selection field, which may be useful for large proteins or to “zoom in” on a given set of residues (FigureB). Adjacent to the threshold sliders, there are buttons to restore each field to its default value: 4 and the equivalent of 0.4 on the selected scale for the length and hydropathy thresholds, respectively (FigureD,E). Additionally, in the viewer window, users are able to adaptively tune parameters and view how this changes the blobs found on a protein’s structure. Finally, using a dropdown menu (FigureF), users can color blobs by their type (all h-blobs are colored the same shade of blue) or by their ID (all h-blobs are colored distinct shades of blue). Scripts and installation instructions for the VMD plugin can be found in the blobulator GitHub repository: https://github.com/BranniganLab/blobulator.

Screenshot of the VMD blobulator plugin. The user sets the Mol ID of the protein (A), provides an atomselection for the display (B), selects a hydropathy scale and may select the auto update option to snap the threshold to a scale-dependent default (C), sets the minimum blob length L min (D) and hydropathy cutoff H (E), and selects how blobs are colored (F). The “Blobulate” button initiates blobulation and creates graphical representations within VMD (G). All representations are deleted by clicking the “Clear representations” button (H). The “Graphical Representations” control panel contains the created graphical representations, which can be further modified (I), and are eventually displayed in the example viewer window (J) showing Leptin (PDB: 1ax8).*

Example Application: Lysozyme

4

In this section, we present the blobulation of lysozyme and demonstrate how blobulation can provide a hydrophobicity-based framework for identifying protein modularity using only the sequence. We have chosen an example protein with previously established features, such as binding sites and disease-associated mutations, and illustrate the context that blobulation provides to these known features and how future research might utilize the blobulator toolkit for proteins with less established features. Additional example applications are presented in Text S4.

In an aqueous environment, most globular proteins have a highly hydrophobic core surrounded by a solvent-accessible surface. One such protein is lysozyme, which cleaves the sugar and peptide components of the peptidoglycan. To provide an example of how one might detect hydrophobic blobs that correspond to the hydrophobic core of a globular protein, we blobulated bacteriophage T4 lysozyme and varied both the hydropathy threshold (H*) and minimum length (L min) (Figure). Lysozyme has two catalytic residues near the N-terminus and several substrate contact sites along the sequence (FigureA). FigureB shows a secondary structure representation of lysozyme with residues colored by blob type, and we note that h-blobs do not align with secondary structure elements. Higher hydropathy and length thresholds eliminate the h-blobs that are detected at the surface of the protein when using more relaxed settings (FigureC). Blobulation using the most stringent settings shown here (H* = 0.5, L min = 8) reveals two h-blobs at the center of the protein, away from the solvent-accessible surface. By gradually increasing parameter thresholds, we can isolate the components of a globular protein that correspond to the core as well as the shorter and less hydrophobic blobs that interact at the surface of the protein. This is consistent with previous findings that h-blobs tend to be buried in structured proteins.?

Blobulation of lysozyme. (A) Blobs colored according to the blob type, as outputted from the blobulator webtool, and produced using default settings (H = 0.4, L min = 4). Annotations indicate catalytic residues (red) and the substrate binding site (purple). (B) Molecular image of lysozyme (PDB: 148L) with peptidoglycan (green) colored by the substrate binding site (left, residues found within 7 Å of peptidoglycan) or by the blob type (right, h-blob: blue, p-blob: orange, s-blob: green). (C) Blobulation under increasingly stringent settings (H* = 0.4 and 0.5, L min = 4 and 8). H-blobs are shown as surfaces. Molecular images were generated in VMD , using the VMD plugin introduced in 3.3.*

To investigate whether the substrate binding site is composed of h-blobs, we identified blobs containing contact residues within 7 Å of the peptidoglycan (shown in Figure). When using a shorter L min (L min = 4), h-blobs are found contacting the peptidoglycan across its entire length. However, when using a longer L min (L min = 8), the only detected h-blobs in the substrate binding site are those in contact with the peptidoglycan sugar component. The stabilization of this sugar ring is vital for the ability of the enzyme to cleave the peptidoglycan,? and we find this ring wedged between two long hydrophobic blobs (FigureC), providing an example of long hydrophobic blobs with interactions critical for function.

Blobulating lysozyme using “relaxed” settings (H* = 0.4 and L min = 4) reveals two examples of “blob groups”: sets of h-blobs separated only by s-blobs (as defined in section, shown in gray in FigureA). Both blob groups are found near the center of lysozyme, surrounded by ungrouped h-blobs and p-blobs. Additionally, each group contains an h-blob that remains detected under increasingly stringent settings and is also found at the core of lysozyme (FigureB, shown also in the molecular image in ?). The groups detected here are akin to tertiary elements formed from a network of interacting secondary structure elements, such as α-helical bundles. This is an example of hierarchical blob clustering, where hydrophobic blobs detected at restrictive parameters are found in the protein core surrounded by less hydrophobic blobs within the same group, which are in turn surrounded by individual h-blobs.

Blob groups in T4 lysozyme. (A) Structural view (PDB: 2LZM) blobulated using default settings (H = 0.4, L min = 4). Groups are gray; ungrouped h-blobs are blue, and p-blobs are orange ribbons. (B) Blobulation of T4 lysozyme under increasingly stringent settings. Annotations indicate blob identifiers and blob groups. Molecular images were generated in VMD. ,*

To provide an example of how blobulation links known mutations with an effect on their local hydrophobic context, we blobulated the lysozyme mutants S117V and T157I. Both mutations alter the effect of temperature on the stability of the lysozyme by altering intraprotein hydrophobic interactions. S117 makes the protein more stable at higher temperatures by altering hydrophobic residue interactions in the substrate binding cleft,? while T157I makes the protein less stable at higher temperatures and disrupts hydrogen bonding at the periphery of the protein.? We find that both mutations change the blob topology (Figure): S117V merges two h-blobs, and T157I creates a new h-blob four residues in length. We have previously found that mutations that split, dissolve, or merge h-blobs are enriched for deleterious mutations,? and this result is consistent with that finding. Additionally, the h-blob introduced by the T157I mutation is classified as a Janus region on the Das–Pappu phase diagram.? Janus proteins often have degenerate conformations and switch between ordered and disordered depending on their environment,? which may cause a change in this blob at higher temperatures and lead to less overall stability for the protein. Finally, single-residue mutations frequently do not cause detectable structural differences in predictions; for instance, AlphaFold predicts minimal structural differences between these mutants and the wild type (RMSD < 0.1 Å). In contrast, the sequence-based topology from blobulation is sensitive to these single-residue mutations.

Globular tendency of T4 lysozyme blobs. Blobulation using default parameters (H = 0.4, L min = 4). Blobs for each sequence are colored by Das–Pappu phase, as in Figure I. Red diamonds indicate mutated residues. S117 V, which increases thermostability, joins two h-blobs into one. T157I, which decreases thermostability, creates a new h-blob (as well as an s-blob).*

Conclusion

5

Here, we presented the blobulator toolkit for characterizing and visualizing patterns of contiguous hydrophobicity in proteins. Blobulation reflects higher-level sequence organization, much like secondary structure elements, but can be applied even when structural data are unavailable. The runtime of the CLI is comparable to that of modern secondary structure predictors and scales about four times more efficiently with sequence length. The webtool and VMD viewer provide a graphical and interactive means to explore the hydrophobicity in a single protein in depth, including blob-level properties like net charge and disorder, as well as the impact of amino acid substitutions.

As illustrated in the example applications and in ref ?, sequence partitioning by hydrophobic blobs is compatible with known functional segments and tertiary interactions. Tertiary interactions are increasingly incorporated into bioinformatics analyses ?,?−? ? ? ? ? ? ? typically through machine-learning approaches. Since hydrophobic blobs are critical determinants of tertiary interactions, the blobulator CLI suggests a deterministic, lightweight, and biophysically informed route to bioinformatics analyses. For example, this toolkit opens the door to studies that require high statistical power across multiple sequences, such as the long-term evolution of protein hydrophobicity or the impact of disrupting hydrophobic blobs for genetic disease.

Supplementary Material