SMLG (Statistical Machine Learning Group) Discussion Forum

Query Series GSE198411 (RNAseq)

Experimental condition (Filename)
===========================================
control #1 (h-cont-1.txt)
control #2 (h-cont-2.txt)
shRNA67 #1 (h-sh67-1.txt)
shRNA67 #2 (h-sh67-2.txt)
shRNA103 #1 (h-sh103-1.txt)
shRNA103 #2 (h-sh103-2.txt)

Each tab delimited text file contains Transcript ID, Gene Symbol, and raw counts.

Genes of interest:

TFEB. TFE3, LAMP1, LAMP2, HSPA8, SLC39A14, VAMP7, TRPML1, SNAP23, GABARAPL1, ATP6V0E1, ATP6V0D1, ATP6V1G1, ATP6V0C, ATP6V0A4, cathepsin D, cathepsin L, cathepsin S, cathepsin H, MT1E, SQSTM1, MAP1LC3B, ATG7, LMNA, LMNB, LMNB2, CLDN7, CLDN1, CLDN12, and CDKN1A.

cwyoo wrote:Query Series GSE198411 (RNAseq)

Experimental condition (Filename)
===========================================
control #1 (h-cont-1.txt)
control #2 (h-cont-2.txt)
shRNA67 #1 (h-sh67-1.txt)
shRNA67 #2 (h-sh67-2.txt)
shRNA103 #1 (h-sh103-1.txt)
shRNA103 #2 (h-sh103-2.txt)

Each tab delimited text file contains Transcript ID, Gene Symbol, and raw counts.

Genes of interest:

TFEB. TFE3, LAMP1, LAMP2, HSPA8, SLC39A14, VAMP7, TRPML1, SNAP23, GABARAPL1, ATP6V0E1, ATP6V0D1, ATP6V1G1, ATP6V0C, ATP6V0A4, cathepsin D, cathepsin L, cathepsin S, cathepsin H, MT1E, SQSTM1, MAP1LC3B, ATG7, LMNA, LMNB, LMNB2, CLDN7, CLDN1, CLDN12, and CDKN1A.

Create one dataset that combines all the above six tab delimited files (referred as "six files") by the following steps:
Step 1. Find out the sets of genes that are common (referred as "common genes") in all six files
Step 2. For each gene in the common genes, record each raw counts for each of six experiments
Step 3. For each raw counts, perform "DESeq2-normalized counts: Median of ratios method" presented in https://hbctraining.github.io/DGE_workshop_salmon/lessons/02_DGE_count_normalization.html
Step 4. Create a variable called "ZIP11 Knock Down" and assign "0" for two controls, "1" for two shRNA67, and "2" for two shRNA103.
Step 5. Using the dataset file from Step 4, create a dataset with roughly 200 genes that show the highest fold change in controls compared to shRNA67 (roughly 100 genes) and controls compared to shRNA103 (roughly 100 genes). Add expression levels of Genes of Interest and include "ZIP11 Knock Down" variable in this file.
Step 6. Using the dataset file from Step 5, discretize each gene's expression to 0, 1, 2 using Z-score.
Step 7. Perform banjo analysis on the file created in Step 6.

Here are the results of 4 banjo runs (1h, 4h, 16h, 32h):
runtime | server | status | score
-----------------------------------------
1 hour | path6 | -529.3695
4 hours | path4 | -528.0048
16 hours | path3 | -528.9517
32 hours | path5 | -526.9734


I plan to update this post with highlighted genes of interest and parents/children of ZIP11_KnockDown shortly.

Dr. Yoo, should I also generate a sub-network in a similar fashion as the mouse liver experiment network?

Update: included highlighted genes of interest (green), parents that are not genes of interest (light blue); ZIP11_KnockDown (light yellow) and children that are not genes of interest (red)

samanthagonzales wrote:Here are the results of 4 banjo runs (1h, 4h, 16h, 32h):
runtime | server | status | score
-----------------------------------------
1 hour | path6 | -529.3695
4 hours | path4 | -528.0048
16 hours | path3 | -528.9517
32 hours | path5 | -526.9734


I plan to update this post with highlighted genes of interest and parents/children of ZIP11_KnockDown shortly.

Dr. Yoo, should I also generate a sub-network in a similar fashion as the mouse liver experiment network?

Update: included highlighted genes of interest (green), parents that are not genes of interest (light blue); ZIP11_KnockDown (light yellow) and children that are not genes of interest (red)

Yes, please generate a sub-network in a similar fashion to the mouse liver experiment network. Thanks,

Here is the subnetwork. Highlights in the dot-generated graphic are as follows:

yellow: markov blanket of ZIP11KnockDown

green: genes of interest within 3 genes of ZIP11KnockDown

purple: parents of the children of ZIP11KnockDown

red: children of ZIP11KnockDown

white: genes necessary to retain a connected network

Let me know if you have any questions