Weighted correlation network analysis (WGCNA) tutorial in R| PART I

[@drgutharajasekar6275]

why Module eigenegene is called first conponent of PCA and why this First component is required in WGCNA. Average gene expression is of a module is not enough for ME caluculation.

[@grsbiosciences]

I don't have many traits like weight , cholesterol to correlate. I have only normal and disease traits. What I can try with wgcna

[@dr.ragill2752]

Please make a video on "Cytoscaping of a significant module's edge and node", next step is from here. Thanks!

[@LiquidBrain]

Hi thank you, I have made a video about cytoscape and rcy3 package, perhaps it’s what you wanted kzbin.info/www/bejne/hJSpoZ-Yjb56iNU

[@dr.ragill2752]

@@LiquidBrain Thank you for your reply. I will try. Can you please share your email?

[@saraalidadiani5881]

Thanks for the nice turorial! may you please explain what is the difference between signed and signed hybrid in Network type? I couldn't get it and when and based on which question I have to choose Signed or signed hybrid? thanks :)

[@siefeldinsobih4328]

one provides negative values (signed), and one gives you absolute values (unsigned). If you're dealing with biological data and want to see negative correlation you might want to use signed.

[@bobby5625]

Hi! Beginner here. I am using mRNA seq V2 data normalized using RSEM. I got my data from TCGA. Can I still use the steps you provided in the video? I tried the sample clustering, but I did not get comprehensive results. No clustering was done. Can you provide your insights about this?

[@LiquidBrain]

Yea, I think it’s ok to use TCGA data for WGCNA, what is the phenotype data you use for the correlation? My understanding is that WGCNA allows us to group the genes of the sample into different modules, based on their respective correlation score of their discrete or continuous phenotypes.

[@RakeshBhowmick]

Is it necessary to take 5000 samples or you can directly conduct analysis on whole dataset

[@LiquidBrain]

Hi there, yes 5k samples was for demonstration, should use the whole dataset :) -Lind

[@RakeshBhowmick]

@@LiquidBrain thank you. I could make dendrogram by automatic method. But during TOM calculation step 2 it keep on running without any output. Sometimes showing error

[@howcani8328]

Thank you so much for your valuable video. I have a question, What do you mean by Jintao data?.... Can you please make video for Data preprocessing as a preparation for WGCNA?

[@LiquidBrain]

Hi sorry for my accents, it’s the “gene count data” that can be downloaded from the GitHub link XD The input data is any normalised gene count data, for example, here it’s the Fpkm matrix file - Lind

[@howcani8328]

@@LiquidBrain Thank you so much... your videos are unique and very helpful

[@muhammadanees2008]

Thank you so much for your valuable tutorials, I have a question: In first step we are using gene count table and metadata file to get the fpkm values? if I am using fpkm file instead of gene count and metadata table, then from where I have to start?

[@LiquidBrain]

Hi, glad that you found this useful. Yes you can skip the fpkm normalization and jump right into the WGCNA step.

[@muhammadanees2008]

@@LiquidBrain my bad, am still not getting you :) can mention the line number?

[@paroletatel]

I just wanna thank you a lot.

[@LiquidBrain]

You're welcome, enjoy the videos :)

[@johirislam8174]

I want to analyze WGCNA for my common differential expressed gene ,that I found from 3 different data set.I have the .csv file with the parametter p value,adj.p.value.logFC ,AveExpr,B. value for each of the 3 dataset.And also have the expression value for that three dataset(raw and processed).So now how can I analyze the WGCNA for my DEG.or what type of fdata file is needed to do that??

[@LiquidBrain]

Hi it doesn't start from the log2FC output file. You'll need a table of normalized gene count data, with the sample names in row, genes in column; then you need another table showing the metadata, with sample names as the first column and group ID as the second column. You may refer to my WGCNA github for better understanding of the file structure. github.com/Lindseynicer/WGCNA_tutorial

[@johirislam8174]

@@LiquidBrain in your git hub i didn't find the metadata sample.If i can able to see the sample of metadata than it will be easy for me to interpret. Moreover i want to do WGCNA for my identified DEG.For that reason what should i do.In my sample file my gene name/gene id held in row and the sample id(patients and control) is in the column.So in that orientation of my table,is it possible to do analysis.

[@zarlishattique4167]

@@johirislam8174 same problem

[@prachetaspatel6314]

Could you explain what should I do to input ChIP seq data?

[@LiquidBrain]

Hi sorry for the late reply, I didn't really explore ChIPseq data, but I guess the concept should be similar. To do the analysis, you'll need a table of normalized gene count data (expression values of ChIPseq?), with the sample names in row, genes in column; then you need another table showing the metadata, with sample names as the first column and group ID as the second column.

[@prachetaspatel6314]

@@LiquidBrain Hey, so I have an issue with my data I think since I cannot get a good threshold even if I change parameters for power. I did differential peak analysis with DiffBind, and got a normalized table using EdgeR ( I didn't use DESEQ2 due to the nature of my data). EdgeR normalizes the data differently than DESEQ2, so I am not able to get the same results. I am stuck with a list of genes and an input matrix with normalized peakscores - but the soft power threshold stops at 0.04, instead of reaching 0.9, is there a way to further normalize my data? or should I just use the raw counts with my differential gene list and convert it to FPKM values?

[@rohitbhojane4447]

please can you provide link for sample_info file its urget

[@LiquidBrain]

Hi Rohit, I have just uploaded the sample_info in the same link, hope this helps :) github.com/Lindseynicer/WGCNA_tutorial - Lindsey

[@sankadines]

Hi I want only 1 module and export it to cytoscape format. Can you please help

[@johirislam8174]

Helow I was very nice of you.If you gave me the metadata of this tutorial.From your github page I cannot able to find that.If i can able to found the metadata than i will be easy for me to run my data.But without that it is not possible

[@LiquidBrain]

You can refer to the script here (github.com/Lindseynicer/WGCNA_tutorial) - Brandon

[@charmyshah8944]

Hi thanks for your videos how can I do WCGNA analysis for microarray data?

[@197ankitakumari4]

Your videos are really helpful. Thank you !

[@LiquidBrain]

Glad to hear that!

[@Sebastian-dp9sk]

Your English is sooooo nice.

[@haotingjia5106]

very helpful thanks

[@md.ishtiakrashid1523]

Hi , Thanks for your videos. Is it possible to use 16S amplicon data for wgcna analysis?

[@LiquidBrain]

Hi there, literally it can be used for any genomic data, since it is just an algorithm to cluster the data into different modules. -Lind

[@段绍伟]

Thank you!

[@이윤항-w7d]

Hello Dr. Lian. This is Lee from South Korea. Your video has helped me so much. Thank you so much.

[@LiquidBrain]

So glad to hear about it, thanks

[@이윤항-w7d]

It would be helpful if you provide a script or url for signed analysis, as the default setting deals with unsigned method.