Mar 02, 2018
Functional annotation and analysis of de novo transcriptomes with TRAPID
- Francois Bucchini1,
- Michiel Van Bel1,
- Klaas Vandepoele1
- 1VIB-UGent Center for Plant Systems Biology
- SINGEK_ITN
External link: http://bioinformatics.psb.ugent.be/webtools/trapid
Protocol Citation: Francois Bucchini, Michiel Van Bel, Klaas Vandepoele 2018. Functional annotation and analysis of de novo transcriptomes with TRAPID. protocols.io https://dx.doi.org/10.17504/protocols.io.nhxdb7n
License: This is an open access protocol distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working
Created: February 27, 2018
Last Modified: March 28, 2018
Protocol Integer ID: 10519
Keywords: TRAPID, transcriptomics, functional annotation, comparative genomics, RNA-seq
Abstract
TRAPID is an online tool for the fast, reliable and user-friendly analysis of de novo transcriptomes, developed and maintained by the CNB group at VIB-UGent Center for Plant Systems Biology.
This protocol explains how to perform functional annotation of a transcriptome using the TRAPID platform. It also describes basic analyses that can be performed within TRAPID, after initial processig of the data.
Guidelines
In this protocol, we work with an example dataset consisting in 25,392 transcript sequences from Panicum hallii (Meyer et al. 2012). This dataset can be retrieved from TRAPID's FTP here.
Before start
Make sure that:
- You are using a web browser with JavaScript and Adobe Flash enabled. Supported browsers: Safari, Chrome, Firefox (most recent versions).
- Java is installed on your machine (version JRE 6.0 or higher). This is required to launch the JalView alignment editor and the Archeopteryx tree viewer.
User Authentication
User Authentication
Setting up an account
Before you can use TRAPID, you need to create an account.
Go to the registration page. To go there from TRAPID's homepage, click on the "Register" link at the bottom of the page.
Fill in the required information (valid email address used as login, organizaion and country). Make sure to provide a valid e-mail address as your password will be sent immediately by e-mail.
Click on the "Register" button.
Log in
On TRAPID's homepage, click "Login" at the bottom of the page. This will take you to the login page. There, use the e-mail address used to register and the password sent by mail to login.
After you've logged in, the homepage will be replaced by the Experiments overview. Since you are a new user, this overview will be empty. This page is composed of 3 main sections:
Current experiments (experiments you uploaded and own),
Shared experiments (experiments uploaded by others you are allowed to view),
Add new experiment where a new experiment can be started.
Note
By requiring authentification, we ensure that no user has access to the data of any other user and that each experiment's data is kept private.
Creating experiments
Creating experiments
Create a new experiment
From the Experiments overview page, create a new experiment. To do so, just fill in tour experiment's information in the Add new experiment section, as shown in figure 1. You can for example enter 'Tutorial 1' as a name and 'Panicum transcripts' as description. The reference DB should be left to its default setting (PLAZA 2.5).
We recommend using PLAZA as reference database when working with plants and algae. If data from other organisms is analyzed, select OrthoMCL-DB 5.0 as reference database.
Figure 1: experiments overview. At the bottom a name and description for the new experiment needs to be filled in for the new experiment. Experiment are created by clicking the Create experiment button.
Add the experiment by clicking Create experiment. The new experiment will now appear in the current experiments list. Note that each user can have a maximum of 10 experiments.
Accessing your experiment and experiment overview
Click on the experiment's name in the Current experiments table to continue. This will take you to the experiment page.
From there, general statistics are shown (Experiment information), sequences can be imported and exported (Import/Export) and a search function is available to find specific sequences (Search). There also are options to start processing transcripts (Initial processing, more details in section 4). Once data is imported, a detailed overview of the transcripts and a toolbox will also become available.
An example of experiment page is given in figure 2.
Figure 2: experiment page. The experiment page of our currently empty experiment. Click on 'Import data' to import transcript sequences.
Note
In this protocol, we work with an example dataset consisting in 25,392 transcript sequences from Panicum hallii (Meyer et al. 2012). This dataset can be retrieved from TRAPID's FTP here.
Uploading transcript sequences & job management
Uploading transcript sequences & job management
From the experiment page, click on Import data to go to the data upload page (referred as Transcript file management page in TRAPID). Two options are available:
- in case you downloaded the sample Panicum dataset, you can upload the file (use Browse to locate the file on your system),
Loading the data into TRAPID is then done in two steps. First, you need to upload the dataset (or datasets in case you are using multiple input files) by clicking Import transcript sequences. Then, click Load data from files into database to get the sequences into our database. Both steps are essential before the data can be processed. A screenshot of the transcript file management page is shown in figure 3.
Figure 3: transcript file management page. From this page, transcriptome dataset(s) can be uploaded from your machine or a URL. After the data file(s) were added to our system, click the Load data from files into database button.
Once your data was successfully uploaded to TRAPID, an email will be sent to your login email address.
During all TRAPID processing steps (upload, transcript processing, running frameshift correction or computing alignment/phylogenetic tree), you can check the experiment status to see if your job is queued, running or in error status. In case you want to cancel or stop your job, go to the Experiment Status page (accessible form the experiments overview) and modify the status to Finished.
Note
In this protocol, we work with an example dataset consisting in 25,392 transcript sequences from Panicum hallii (Meyer et al. 2012). This dataset can be retrieved from TRAPID's FTP here.
Note
Requirements for input data:
* All data needs to be provided in FASTA format.
* Each sequence has a unique identifier (max. 50 characters) and no empty sequences should be present.
* Single files cannot be bigger than 32Mb (the limit is 300Mb if the file is provided as URL). For large datasets it is possible to use (g)zipped files.
Process transcript sequences (functional annotation)
Process transcript sequences (functional annotation)
After the sequences have been loaded to the TRAPID database (you received the completion email), you can now proceed with the initial processin of your transcripts. To do so, go to the initial processing page by clicking the Perform Initial Processing link on the experiment page. During this step, you should consider the options carefully, as they may seriously impact the subsequent analyses.
On the initial processing page, you have to specify how transcripts should be assigned to gene families and how should functional annotation be performed. For this protocol, you can select the settings shown in figure 4. Finish by clicking the Start transcriptome pipeline button. While an experiment is being processed, it cannot be accessed: the experiment will be available again once the initial processing finished.
Initial processing options
Several settings can be adjusted for initial processing:
- Similarity search database type: whether either a single species, a phylogenetic clade or the gene family representatives will be used for the similarity search (first step of TRAPID's initial processing). A single species is a good choice if in the reference database a close relative of the transcriptome species is present. If a good encompassing phylogenetic clade is available, then this is also a solid choice. If none of the above, then the gene family representatives will provide a good sample distribution of the gene content within each reference database.
- Similarity search database: depending on the choice made for the similarity search database type, this list offers you choices of database to select.
- Similarity search e-value: set an E-value cutoff for the RapSearch2 similarity search.
- Gene family type: define the Gene Family Type. For PLAZA 2.5 this is Gene families (TribeMCL clusters) or Integrative Orthology (only available if a single species was selected as database type), for OrthoMCL-DB this is Gene families (OrthoMCL clusters).
- Functional annotation: define how the functional annotation should be transferred from the family to the transcript level. In general, transfer based on gene family is the most conservative approach while transfer based on best hit is yielding a larger number of functionally annotated transcripts. Logically, combining both methods using transfer from both GF and best hit yields the largest fraction of annotated transcripts.
Figure 4: initial processing page. On this page, you can select multiple options regarding how the sequences should be added to gene families and how to perform funtional annotation.
Once the initial processing is done, all sequences will be included in gene families and will be functionally annotated, if possible. Additionally, as transcript data often includes truncated sequences or sequences with indels, problematic sequences are flagged, for example if they contain a putative frameshift.
You will a receive an email when the processing is finished.
Note
Initial processing is rather fast. A test data set containing 90,000 transcripts can be processed in less than 3 hours, with approximately 28% of these transcripts assigned to gene families. For our example Panicum data set (25,392 transcripts), the complete processing, including upload and transcript processing (using Monocots clade) takes around 1 hour (and yields 60% of transcripts assigned to gene families). Note that the fraction of very short or very long transcripts will impact the total processing time during this initial phase.
Exploring TRAPID output: basic analyses
Exploring TRAPID output: basic analyses
General statistics
From the experiment page, use the General statistics link in the toolbox to access the general statistics page (example shown in figure 5). This page offers a complete overview of ORF finding, gene family assignments, similarity search species information, meta-annotation and functional information.
Figure 5: general statistics page.
Subsets and labels
If the dataset is comprised of transcriptome data from different sources (corresponding for example to different tissues of a multicellular organism, or different stress conditions), then it is possible to assign labels to the subsets.
By doing so, it becomes easy to browse the function of a subset of transcripts, and several new analyses become available: comparison of functional annotation, and functional enrichment analysis. The latter is covered with more details in our other protocol (see section 10).
To define a subset, click the Import transcript labels/Import data link in the import/export data section of the experiment page. You then have to provide, as file, a list of transcripts that should be form the new transcript subset. Note that a transcipt can be in multiple subsets.
You could for example define a new subset using a set of Cell Cycle genes for our Panicum dataset. To do so, hit Browse to select the file, enter a label name (e.g. Cell_cycle) and click Import labels.
Searching for data
From the experiment page, use the Search box to search for a number of possible data types within the current experiment.
Relevant families can be found using the search function, for example by looking for a protein domain / GO term of interest and browsing the list of associated sequences.
Note
Functional annotation can be searched for both through direct term identifiers (e.g. 'GO:0005509') or through the descriptions (e.g. 'Calcium ion binding').
Exporting data
TRAPID enables users to export both the original data and the annotated/processed data of an experiment. To access the export page, click the Export data link in the import/export data section of the experiment page.
It is possible to export:
- Structural data: ORF information, transcript/ORF/protein sequences
- Gene family data: transcripts with their associated gene family, gene families with their associated transcript, gene content of the reference gene families
- Functional data: GO & InterPro information (annotation and meta-data).
- Transcript subsets, in case you defined any (list of identifiers)
Data export is performed in two steps: first, click on the button corresponding to the data type you want to export (for example transcripts with GF), then click on the generated download link (download file). This two-step implementation is due to the fact that files can take a while to be generated.
Note
The export of the functional GO information has an extra column 'is_hidden', indicating whether a GO term is flagged as hidden, due to the presence of more informative GO terms in the GO graph for the given transcript.
The toolbox
On most pages (experiment/transcript/gene family/GO/protein domain), a toolbox is available which contains the most common analyses to be performed on the given data object.
For example, click on a gene family in the transcript table of the experiment page. From there, you would be able to create multiple sequence alignments or phylogenetic trees, in the gene family context, for transcripts associated to this gene family, using the options available in the toolbox.
Going beyond
Going beyond
Although this protocol provides a good introduction to TRAPID and its basic functionality, it is possible to conduct more advanced functional and comparative analyses on the platform. Please read our other protocol, 'In-depth functional and comparative analyses of transcriptomes with TRAPID', for more details.