DRAW: tool for RNA-seq analysis

Designer for RNA-seq Analysis Workflow. DRAW is a tool that builds workflows from predefined components on the basis of parameters provided by the user from the drop-down lis. It then creates a configuration file, thus enabling repeatability of the analysis and the documentation its course.

The tool enables different solutions depending on whether the data needs preprocessing. To make it easier for the user to use many different, we placed our tool in Docker. Then, very time-consuming software installations and hardware requirements are easily avoided.

The final stage of the analysis is the dashboard visualisation.

Input data requirements

Input data should be placed in input directory. It should contain:

• single end or paired end .fastq files,
• annotation file in .gft format,
• reference sequence in .fasta format.

WARNING DRAW is able to process only .fastq files. When using paired end files it is is necessary to label corresponding paired end files should with _1.fastq and _2.fastq.

Isolated environment in Docker containers

Before starting the analysis make sure you have installed Docker (recommended version 19.03.8) and Google Chrome (recommended version 81.0.4044.113) on your Linux machine. It is the only installation needed, the rest does Docker itself.

To install Docker on Linux follow all instructions from website linked below, including setting up the Docker Group to avoid "Permission denied" issue.

https://runnable.com/docker/install-docker-on-linux

Preparing the environment can be done only once, Docker image once build exist on Your machine and can be accessed at any time. Environment requires 2.1739 GB of Your disc space. It takes approximately 40 min, during image building process all needed platforms, tools and requirements are installed. To get the image of DRAW, make sure You have Internet access and execute the following script.

$ ./build.sh

Before running the container place input data in the input directory. To run the container execute the run_docker.sh script as below. Directories input, config_files and output persist data generated and used by Docker. Ports 2000 i 5000 are published to the host machine. Container is running in the interactive mode for user to see running processes. Execution of the script will open a GUI application to fill in configurations and visualizations server, which will rise after the end of the analysis.

$ ./run_docker.sh

GUI and visualizations are available while Docker container is running, you can simply acces them typing URLs in browser as below. GUI is accessed by port 2000, visualizations by port 5000. Mind visualizations are available after generating config file and running analysis, which takes some time.

http://0.0.0.0:2000/index

http://0.0.0.0:5000/bkapp

To stop the docker from running use Ctrl+C

If encounter an error "Adress is already in use" type following command and port, which you want to release:

$ sudo fuser -k Port_Number/tcp

Docker uses following options:
--tag, -t Tag docker container.
--publish , -p Publish a container’s port(s) to the host
--volume , -v Bind a volume
--interactive, -it Run container in interactive mode

Graphic User Interface (GUI)

This simple flask app offers limited functionality that allows basic config file management for potential user. GUI opens in default browser installed on user's web browser. User is greeted with short software description and two options: "Generate", "Load config".

Generate - two-step config file generation procedure. During first step user is obliged to specify paths to input files, run type and other details regarding inputs. Default run_id consists partly of current computer time to maintain config files' uniqueness. Second step varies depending on which run type was specified in previous step. By filing second form user is able to define all necessary parameters for tools used in analysis. Hovering cursor over specific parameters prompts tooltips that give some insight into their effect on given tool. Subsequently config file, which name is unique run_id, is created in form of json file in config_files directory located in main folder. At this point user is presented with an option to perform run using config paramteres that were just specified.

Load config - this option prompts user for run_id of previously created json config file. That way user can repeat analysis with previously specified conditions and tools.

It is crucial for user to specify parameters for all chosen tools. This allows for further, successful analysis

Differential expression analysis

The differential expression analysis is performed in DESeq2. The output of the differential expression analysis are 4 files:

• Normalized counts
• rLog - regularized logarithm
• VST - variance stabilizing transformations
• DESeq2 result file

Visualisation

The visualisation dashboard offers four type of plots, which can be accesed by selection appriopriate tab. Most often the visualisation dashboard will be opened as soon as the analysis is finished. By default, the data used for presentation of the results is variance stabilizing transformation, VST.

When needed, the visualisation for a given run can be accessed in a independent manner. All of the requirements for running the analysis apart from the Docker and the rest od the DRAW module, can be installed:

$ ./vis_requirements.sh

Then, to run the visualisation:

$  cd bin # Make sure that you are in bin folder!
$  python3 vis.py -id [run_id] -dt [data]

where run_id is the id of the run we want to visualise, and data is one of the followind data types:

• norm - normalized counts
• rlog - regularized logarithm
• vst - variance stabilizing transformation

The plots offered:

Smear plot

Shows the the average log2 count per million (CPM, x-axis) versus log2 fold-change (FC, y-axis) of the change in gene expression.

Text box allows choosing FDR cut-off. Red points indicate genes found to be significantly upregulated or downregulated by the treatment. All genes are labelled and their details can be accessed by moving the cursor over a gene of interest.

Volcano plot

Volcano plot is a scatterplot showing statistical significance (p-value) versus magnitude of change (FC, fold-change). It enables quick visual identification of genes with large fold changes that are also statistically significant.

Thanks to text input it is possible to select p-value cut-off according to which the non-significant genes are separated from the sigificant ones, by colouring them in grey. The colour of significant genes in controlled by LogFC parameter. Text input allows for choosing the threshold by which the user decides which genes are considered up- (colored in red) or down-regulated (colored in blue). All genes are labelled and their details can be accessed by moving the cursor over a gene of interest.

Heatmap

Heatmap is prepared using pairwise_distances from sklearn library and dendrogram and linkage from scipy. The dendrogram at the side shows us a hierarchical clustering of the samples.

FDR cut-off enables narrowing the clustered genes to a selection of choice. Additionally, the user can further subselect the genes, choosing the number of genes to plot with a given FDR. All genes are labelled and their details can be accessed by moving the cursor over a gene of interest.

PCA

Principal component analysis is performed and two principal compontents can be selected by user for plotting from drop-down menu.