Sep9 2019 Meeting - quevedor2/schramek_wiki GitHub Wiki
In attendance:
- Samah, Rene, Ellen, Daniel, Trevor
Goal:
- Initial meeting with Trevor to outline the deliverables for the project.
Schramek Project
ATACseq
Notes:
-
Need to include an ENCODE mouse ATAC sample as a negative control
-
Scrambled: different biological tumours, so entirely separate
- Same scrambled guide and sgRNA
- Makes sense that they are very dissimilar
- All of the mouse samples are from separate tumors, but of the same strain of mouse
-
Check if the dissimilarity is due to the number of reads difference
-
One of the APC samples, the RNA had very poor quality
- Did not make the cutoff for RNAseq, but was just submitted for ATAC
- We should do Picard QC metrics on these samples as added on QC, might explain some dissimilarity
- Did not make the cutoff for RNAseq, but was just submitted for ATAC
-
Take the overlap (recursively) and then cluster
- Find the overlap between replicates and cluster based on those peaks
- Find the overlaps between samples, and cluster based on those peaks
- There may be a way to create a saturation curve here....
-
Promoter peaks are a bit low
- Samah: "Coding-exons are depleted in ATAC-seq, the TN5 favors more promoter regions rather than coding regions"
- To see whether this is a thing, we need to make a distribution plot divided by the genomic footprint (normalize for it)
- Promoter annotations for mouse versus human may be different. We need to standardize between the two species to make the comparison between plots better. Maybe take 3k bp upstream of TSS for both humans and mouse?
- Polycomb goes up, enhancer should go down? MLL is involved in this
- Samah: "Coding-exons are depleted in ATAC-seq, the TN5 favors more promoter regions rather than coding regions"
-
Encode mouse for genotype calling
-
Run CREAM for super-enhancer calling
-
Picard QC metrics for the bams
Questions to address:
- Are enhancers differentially regulated?
- If we see differences in peak distribution, do we see distribution difference in enhancers or LINE elements?
- Calling of super-enhancers
- RNA-seq: Map from open-chromatin to differentially regulated genes
- Nominate enhancers and link them to certain genes using C3D
- Using RNA-seq they see clear clusters for PC clustering. If we do the same with ATAC-seq, can we reproduce those clusters?
- Comparison with ENCODE SCREEN data, we need to standardize pipeline:
- candidate cRE's with the enhancers that are lost or gained from the mouse models
Data needed
- Ellen needs to transfer the Asxl Rep2 mouse ATACseq data
- Ellen needs to transfer the RNAseq data. Replicate their clustering.
RNAseq
Ellen did:
- GSEA: de-regulation of ER signalling, increase in gamma and alpha, might be endogenous retrovirus activated
- 30-40 million reads RNAseq for virus calling may not be enough (Nabado paper?)
Samah Project
-
Look for regions that are conserved in promoter regions of the mutation sites
- Motif analysis
-
HNSCC check for the PIK3CA for the promoters versus coding regions
- Look pancancer, pancancer mutual exclusivity
- Differentials between the tumors
- Do a motif analysis on the promoter regions, enhancer or repressors motif LOOK FOR REPRESSIVE MOTIF
- If it's a knockout, much much easier and quicker ot validate
- Loss of a repressor, gain of function on pIK3CA