README.md

In sportiellomike/fluximplied2: Pathway analysis software utilizing rate limiting steps

fluximplied

These R functions assist in the generation of hypotheses for flux analysis from transciptomic or metabalomic data. Instructions for installation and use of those functions are detailed below.

Authors, contributors, and more

• Mike Sportiello MS, Adam Geber, Rohith Palli PhD, Aizan Embong MS, Nate Laniewski, Emma C Reilly PhD, Kris Lambert Emo, and Dave Topham PhD
• This tool was created by students and employees at the University of Rochester Medical Center. We acknowledge and are grateful for the public institutions that support this work as it would have been impossible without them, including the National Institute of Health, our public education system, and more.
• You can contact us at [e-mail].
• We would appreciate you citing us. An acceptable citation follows: Fluximplied: A novel approach integrates rate limiting steps and differential expression for pathway analysis. Mike Sportiello, Adam Geber, Rohith Palli, A Karim Embong, Nathan G. Laniewski, Emma C. Reilly, Kris Lambert Emo, David J. Topham. bioRxiv 2022.03.15.484417; doi: https://doi.org/10.1101/2022.03.15.484417

The problem

Differential expression analysis is increasingly commonplace. For us T cell immunologists, we might look at the number of genes that are upregulated in activated T cells compared to their non-activated counterparts and find that many in our list of upregulated genes have to do with killing target cells (an essential and well established role of CD8 T cells). We might see that, after performing a statistical test using methods such as enrichr, that genes associated with cell killing are enriched in the upregulated gene set.

But here's the problem: what if instead of looking at the whole gene set, we just wanted to look at metabolic pathways and get a sense of which metabolic pathways are upregulated? Let's take glycolysis as an example: if all of the genes of glycolysis are upregulated in activated T cells as compared with non-activated T cells, we would find this to be an upregulated pathway in activated T cells. But, what if only one gene was upregulated? Likely, the p value of the enrichment analysis would be non-significant and appropriately well over 0.05. This might sound good, but what if I told you that one gene was the transcript for the enzyme Phosphofructokinase 1, the rate limiting step of glycolysis? Non-biased gene set enrichment analyses don't factor in flux at all, nor do they take into special consideration the rate limiting steps of these metabolic reactions eithre.

I think most of us would agree that higher levels of the enzyme that is the rate limiting step, the step that controls how much glycolysis is actually going to happen, is meaningful. Imagine a carrot cake assembly line with jobs that all take one minute to do, except for one person in the middle (they're the ones doing the time-consuming job of actually baking the cake), whose job takes 1 hour. Increasing the people around that one person probably wouldn't do much, but increasing the number of people/ovens that do that one rate limiting step would likely greatly increase the speed of the carrot cake assembly line as a whole. And more to the point, it would increase the flux of carrot cakes through the assembly line. The same can be said for many metabolic reactions. The upregulation of Phosphofructokinase 1 implies increased flux through glycolysis, just as an extra worker for the rate limiting step on the assembly line would imply increased flux through the assembly line.

The solution

Here, we present the R package fluximplied. Fluximplied requires a list of genes or a dataframe result from differential expression analysis as the input data, the species that the genes are from (it currently accepts only mouse and human), as well as how those genes are encoded (as official gene symbols or official ENTREZIDs). We have created a publicly accessible database of rate limiting steps and the gene that encodes them. The function compares your supplied gene list with this database and looks for overlaps, and then returns those overlaps to help you generate hypotheses and followup analyses with flux balance analysis or other functional assays.

In order to maintain statisitcal rigor, P values, what we term "P adjust adjust" or "Padjadj", are determined with standard P adjustment of the P values which were adjusted during the differential expression analysis. Fluximplied will adjust the Padj using the Benjamini-Hochberg (BH) method for the number of comparisons being made with the Rate Limiting Step Database (that is, the number of genes within that database).

How to install

Mac and PC

You can use an interactive graphic user interface (GUI) which allows you to upload your CSV (which you can make by exporting from excel), and returns a table of rate limiting steps, a bar graph of your results, and text explaining the meaning of your results. You can access that GUI here: https://sportiellomike.shinyapps.io/fluximplied/

Or, for those with more experience using R, simply install the package (ensure you have an internet connection):

install.packages('devtools')\ library(devtools)\ install_github('sportiellomike/fluximplied',build_vignettes=T)\ library(fluximplied)

Linux

Root access is required. Linux requires certain dependencies installed on your actual machine (not just other R packages). We made this easy on you by supplying a bash script. The bash script is called "install_for_ubuntu-fluximplied.sh". To run it, download it from https://github.com/sportiellomike/fluximplieddev and run it by typing the following into the terminal while being in the correct directory:

bash install_for_ubuntu-fluximplied.sh

We tested fluximplied installation with that script on Ubuntu version: 20.04 LTS (amd64), XFCE4 Desktop Environment; R version tested on ubuntu: v3.6.3. If those installs are not working Navigate to the functions at https://github.com/sportiellomike/fluximplied/tree/master/R, and download them and run those scripts.

We offer all this code for free and in open source, and only ask in return that you cite us so other people can learn about this and help contribute.

How to use

The function is formatted as follows:

fluximplied(inputdat,species='Mmu',geneformat='Symbol',inputformat='df',padjcolname='adj_pvalue',pcutoff=0.05) inputdat may be either a dataframe or a vector of characters. * If the user supplies a vector, there is no method to determine statistical significance of overlap with our rate limiting step database, but the function can still return a table of the overlapped genes. * If the user supplies a dataframe, the dataframe should include, at minimum, genes in the first column, a column that has p values from the differential analysis, and log2foldchange from the differential expression analysis. species may be either Mmu or Hsa (mouse or human). Alternative capitalizations of Mmu or Hsa may also be accepted. inputformat may be either df (dataframe) or vector. It defaults to assuming you have supplied a dataframe unless you specify vector. Your dataframe should be formatted as described above for inputdat. The padjcolname should be the column specified by the user as the Padj column from differential expression. It may default to 'adj_pvalue' unless the user supplies another name. This is the name that DESeq2 uses when the user saves the DESeq2 result as a csv. * pcutoff may be any number between 0 and 1. It defaults to 0.05. This is the alpha the user sets. To be clear, because fluximplied adjusts the Padj, this is the alpha for the Padjadj.

The function will save the table as well as the plot to be plotted later as the user wishes. It will also save the subsetted table from the database for which there is overlap.

Some notes

This software is not currently compatible with other formats beyond Gene Symbols or ENTREZIDs (we recommend using the MapIds function from Org.Mm.eg.db / Org.Hs.eg.db and AnnotationDbi to convert it to one of these formats in R, though other options exist). _ _ _ _ / | | () | () | | | || | _ _ | | | | | | | | | \ \/ / | ' _ \| '_ \| | |/ _ \/ _ | | | | | || |> <| | | | | | | |) | | | / (| | || ||_,//__|| || || ./|||_|_,| | | |_|

sportiellomike/fluximplied2 documentation built on Jan. 30, 2024, 12:47 p.m.