Advance negative data generation RNAs not involve in a interaction

[teresa-m]

Idea is to:

• filter data for RNAs that are higly expressed but not involved in an interaction
  • Use mRNA-Seq data produced in same conditions highly expressed RNAs
  • Find expression cutoff for the RNAs
  • Create DB of RNAs involved in RRIs (position, ID, or Biotype)
  • filter so we end up with list or RNAs not involved in RRIs
• RNAs not haveig a RRI find the protein binding profiles
• extract potetioal binding sides by ignoring postions that are bound by proteins.
• check that they have the same binding profile?

The bining profile can be found her for human: https://doi.org/10.1016/j.molcel.2012.05.021

[teresa-m]

Ideas from Rolf. Not sure if I summarized it corretly?

[teresa-m]

https://dorina.mdc-berlin.de/regulators -> get CLIP data from here or from Dominik

[teresa-m]

paper:
RBP coverage: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE38355
mRNA-Seq: https://www.ncbi.nlm.nih.gov/gds?LinkName=biosample_gds&from_uid=997837
... maybe better total RNA-Seq to also get ncRNAs

[teresa-m]

Downloade Data:
full mRNAseq:
Run | Assay Type | AvgSpotLen | Bases | BioProject | BioSample | Bytes | Center Name | Consent | DATASTORE filetype | DATASTORE provider | DATASTORE region | Experiment | GEO_Accession (exp) |Instrument | Library Name | LibraryLayout | LibrarySelection | LibrarySource | Organism | Platform | purification | ReleaseDate | Sample Name | source_name | SRA | Study | Treatment|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|--|
SRR500121 |RNA-Seq | 36 | 1368141804 | PRJNA167851 | SAMN00997837 | 847624134 | GEO | public | "sra fastq" | "gs s3 ncbi" | "gs.US s3.us-east-1 ncbi.public" | SRX149162 | GSM936076 | Illumina Genome Analyzer II | GSM936076: RNAseq_mRNA | SINGLE | cDNA | TRANSCRIPTOMIC | Homo sapiens | ILLUMINA | oligo(dT) | 2012-06-06T00:00:00Z | GSM936076 | HEK293 cell culture | SRP013463| no treatment (mRNA)|

[teresa-m]

Next steps will be to gerate the following position files:

1. expresse positions of mRNA: take total mRNAs -> aligne using RNAstar (build in genome+ gtf of PARIS analysis) -> Blockbuster (find positions of reads over TH) -> end up with list of genomic positions
2. list mRNA interaction postions
3. list of protein profiling positions

Generate potenetal negative mRNA binding sides by filter out all positions of 2 and 3 in one

Next step: How to generate the negative RRIs?

[teresa-m]

Task on how to construct the data set are written here: Generate trainings data using context #25