Tutorial

Contents
    Introduction
    How to Use the Server
    Submitting a Job
    Analysis Results
    Single Sequence Analysis
    Structure Viewer
    Ramachandran Plot
    Multiple Sequence Alignment

Introduction

Welcome to DeepFold-PLM, our state-of-the-art protein structure prediction platform. This tutorial will guide you through the process of using our service, from submitting protein sequences to analyzing the predicted structures.

  • The model predicts protein structures with super fast speed and high accuracy.
  • Input sequences are limited to 1,500 amino acids for optimal processing.
  • All analysis results can be downloaded for further research.
  • The web server grants free access to all users, including commercial users.

How to Use the Server

Our web application follows a straightforward workflow that makes it easy to predict protein structures. Follow these steps to get started with DeepFold-PLM.

1. Register or Sign In

Create an account or sign in to access the full functionality of our platform. While browsing is available to all visitors, you'll need an account to submit prediction jobs and save your results.

2. Navigate to the Home Page

The home page provides quick access to all features. From here, you can start a new prediction, view your job history, or access the tutorial and paper sections.

3. Submit a Prediction Request

Enter your protein sequence in the input field on the home page, then click 'Continue and preview job' to initiate the prediction process. More details on this step are provided in the next section.

4. Monitor Job Progress

After submission, you can track your job's progress in the History section. The status will update in real-time as shown below:

Pending
In Progress
Completed
Failed

5. Analyze Results

Once the job is completed, click on it to view the results. You'll have access to various analysis tools that help you interpret the predicted structure.

Submitting a Job

Let's walk through the process of submitting a protein structure prediction job step by step.

1
Entity 1

After entering your protein sequence, click the "Continue and preview job" button to submit your prediction request. The system will validate your input and create a new job.

Once your job is submitted, you can track its progress in the History section. Completed jobs will have a "View Analysis Page" button that takes you to the results dashboard.

Analysis Results

Once your prediction job is complete, the Analysis page provides comprehensive insights into your protein's structure. The following sections explain the various analysis tools available.

Single Sequence Analysis

The Single Sequence Analysis provides detailed information about your protein's properties based on the primary sequence. This includes Relative Solvent Accessibility (RSA) values and secondary structure predictions for each residue.

Single sequence analysis visualization

The visualization above shows:

  • The protein's amino acid sequence at the top
  • A graph of RSA values in the middle, helping identify exposed vs. buried residues
  • Predicted secondary structures for each residue at the bottom, with different colors and shapes representing α-helices, β-sheets, and loops

Structure Viewer and Distogram

The 3D structure viewer allows you to explore your protein's predicted structure in a fully interactive environment. Alongside it, the distogram visualization provides insights into residue-residue distances.

3D structure viewer and distogram
Key Features:
  • Residues are colored by pLDDT score, providing immediate visual feedback on prediction confidence
  • Interactive controls for rotation, zooming, and different rendering styles
  • The distogram shows pairwise distances between residues, with darker colors indicating closer proximity
  • Clicking on the distogram highlights corresponding residues in the 3D structure

Ramachandran Plot

The Ramachandran plot is a valuable tool for assessing the quality of predicted protein structures. It visualizes the distribution of backbone dihedral angles (φ and ψ) for all residues in the protein.

Ramachandran plot

The Ramachandran plot helps validate the quality of the predicted structure:

  • Clustering of points in the upper left quadrant typically indicates α-helices
  • Points in the upper left quadrant often represent β-sheets
  • The distribution of points in allowed regions (blue shaded areas) suggests a well-formed structure
  • Points in disallowed regions may indicate strained or unrealistic conformations

Multiple Sequence Alignment (MSA) Depth

The MSA depth visualization shows how many aligned sequences cover each position in your query protein. This information is crucial for understanding the evolutionary conservation and reliability of the predicted structure.

MSA depth visualization

Key aspects of the MSA depth visualization:

  • The y-axis shows the number of sequences that align at each position
  • The x-axis represents the residue positions in your query sequence
  • The black line indicates the effective number of sequences (Neff) at each position
  • Color intensity represents sequence identity to the query (orange to white gradient)
  • Regions with higher MSA depth (more aligned sequences) typically result in more reliable structure predictions
  • Positions with low coverage may indicate insertions unique to your sequence or regions of lower prediction confidence