- This is not the originial Hobotnica tool repository. For the original repo and all the instructions on installation and use please refer to https://gitlab.at.ispras.ru/mirachirkova/differential-expression
- The original paper explaining Hobotnica`s maths guts can be found here https://f1000research.com/articles/10-1260
This version of Hobotnica is an edition of the original tool that was created as a semester project in Bioinformatics Institute, Saint-Petersburg, Russia
Supervisors:
- Evgeny Karpulevich
Students:
- Anton Zhelonkin @tony-zhelonkin
- Alexandra Belyaeva @BelyaevaAlex
- Original Hobotnica computes and visualizes gene-level RNA-seq differential expression between two experimental conditions by 5 tools: DESeq, EBSeq, edgeR, limma voom and NOISeq
- Hobotnica also compares results and calculates the best differential expression tool for input data. It also visualizes intersection of subsets
Find the best tool for DE analysis of RNA-seq macrophages data
- Add additional DE tools into Hobotnica docker container: baySeq, DEGseq
- Compare results from different differential expression analysis tools inside Hobotnica
- This eddition adds bayesian differential expression tool baySeq to the already implemented 5 tools: DESeq, EBSeq, edgeR, limma voom and NOISeq
- Visualises the volcano-plot, heatmap based on the results of the baySeq
- Original Hobotnica guts compare baySeq results with the other 5 differential expression tools, outputs all the visualisations for the rest of them and calculates the best tool from the results of all 6
- All the instructions on the installation and use from the original repo at https://gitlab.at.ispras.ru/mirachirkova/differential-expression hold true for this version of Hobotnica entirely
- The original Hobotnica draws a Venn diagramm of intersections between the 5 implemented DE tools
- This version breaks the Venn diagramm. For the purpose of smooth run the code for Venn diagram was silenced
Hobotnica uses matrix of un-normalized counts, table of annotation for each sample and name of directory for output. An example of the input format may be found in the data folder. The data folder contains prepared un-normalised reads from RNA-seq experiment on M0 and M1 macrophages ('RNA_macrophages_reads.txt') and annotation table ('annotation_for_macrophages.txt') The unprepaired reads are available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE163165 The R script written to prepare the aforementioned reads may be found in the data folder under the name 'reads_preparation.Rmd'
Program gives several files for each tool as output. They all are in requested directory.
'de_plots.pdf' and 'heatmaps.pdf' file contains visualisation of 6 differential expression analysis. Expressed genes are signed.
'*_sig.txt' files contain lists of subsets for each tool corresponding chosen signature. Default signature is top-30 most expressed genes. You can manually change it by set "-n" option in run.R. Files names are 'DESeq_sig.txt', 'EBSeq_sig.txt', 'edgeR_sig.txt', 'NOISeq_sig.txt', 'voom_sig.txt' and 'baySeq_sig.txt' (added in this version)
File 'crossing.csv' contains table with information which tool count this gene as expressed.
File 'hobotnica_scores.txt' contains results of Hobotnica computing for each tool.
File 'crossing_with_best.csv' contains results of the best tool and its intersection with other tools results.
The Venn diagramm functionality is disabled in this version.
Using "-s" option in run.R you can save this files:
-
'*_res.rds' files contain differential expression analysis results. They are different for each tool and might be of interest only for users who want to write their own R code with that results. Files names are 'DESeq_res.rds', 'EBSeq_res.rds', 'edgeR_res.rds', 'NOISeq_res.rds', 'voom_res.rds' and 'baySeq_res.rds'.
-
'*_sig.txt.distmatrix' files contain distance matrixes for each '*_sig.txt' subset. Matrixes are ready to be used in Hobotnica. Files names are 'DESeq_sig.txt.distmatrix', 'EBSeq_sig.txt.distmatrix', 'edgeR_sig.txt.distmatrix', 'NOISeq_sig.txt.distmatrix', 'voom_sig.txt.distmatrix' and 'baySeq_sig.txt.distmatrix'.
The easiest pathway is described here
- Download repository
- Move via terminal into the directory of the downloaded repo
- Type in commandline the following lines to install all the necessary files and libraries
$ Rscript install.R
- To run the analysis type in the commandline
$ Rscript run.R countmatrix annotation output
Where countmatrix is matrix of un-normalized counts file name and annotation is table of annotation file name. output is a name of directory for results. The countmatrix and annotation do not have to be in the same folder with the Hobotnica insides The analysis runs smoothly if you explicitly configure the path to the necessary files and the output folder
$ Rscript run.R /your_path/countmatrix.txt /your_path/annotation.txt /your_path/output
Detailed instructions on use and run from Docker are desribed at https://gitlab.at.ispras.ru/mirachirkova/differential-expression
-
install.R
Installs all required R packages, includingbaySeqandsnow, which is required for parallel computations during bayesian analysis.
Example:Rscript install.R -
run.R countmatrix annotation output [-s] [-n top]
Contains all the pipeline. It needs three command arguments - count matrix file name countmatrix, table annotation file name and output directory name output.
Options:
-s allows to save extra information (see output)
-n top allows to set top number of genes in signature.\ See output of run.R in output data paragraph of readme.
Imports functions from source/DESeq.R, source/EBSeq.R, source/edgeR.R, source/NOISeq.R, source/voom.R, source/signatures_utils.R, source/calculate_distmatrix_utils.R, source/baySeq.R (added in this version) \
Other files were not made for executing and contain functions for different parts of the pipelines.
- 'source/baySeq.R' is an entirely new script with all baySeq analysis and volcano-plot visualisations (up & running)
- 'run.R' several lines of code added to implement baySeq under Hobotnica`s pipeline
- source/signatures_utils.R added code to extract top differentially expressed genes from baySeq, created a duplicate heatmap function for the visualisation of baySeq results
- Venn diagram is disabled until further notice
Although the phenotypic variation under the experiment is obvious, the pattern itself is variable. In our case there were almost no intersections with the best tool on top differentially expressed genes identified with other DE tools
According to the result of the Hobotnica analysis limma-voom seems to be the tool of choice in DE analysis of RNA-seq macrophage data
Results of DE tools comparison using Hobotnica expose current problems of existing software in DE analysis.
Each tool has its own specificities in the assumptions on the statistical properties inherent to RNA-seq data.
No consensus as of today exists on the subject. Hobotnica is intended to put some clarity and may be a
helpful instrument in the hands of a thoughtful bioinformatician
- Add more DE tools into Hobotnica
- Add new features: comparing DE tools run on simulated counts data, etc.
- Validate the results with RT-PCR, proteomics analysis data