find_uORFome: Run whole uORFomePipe prediction

In Roleren/uORFomePipe: uORF prediction in R

Run whole uORFomePipe prediction

Description

Steps:
1: Make directory structure for orf finding, create database assign variables and validate input data.
2. Find CAGE transcripts
3. Find uORFs
4. Create database
5. Fill database with NGS and sequence features
6. Train the random forrest model
7. Predict on uORFs
8. Get analysis plots

NOTE: IF it crashes it will continue from the point you quit, so delete the mainPath folder if you want fresh rerun.
Also do not change working directory after you started running, as this might make the program crash

Usage

find_uORFome(
  mainPath,
  organism = organism(df.rfp),
  df.rfp,
  df.rna,
  df.cage,
  startCodons = "ATG|CTG|TTG|GTG|AAG|AGG|ACG|ATC|ATA|ATT",
  stopCodons = "TAA|TAG|TGA",
  mode = "uORF",
  requiredActiveCds = 30,
  max.artificial.length = 100,
  startCodons.cds.allowed = startCodons,
  stopCodons.cds.allowed = stopCodons,
  biomart = "ensembl",
  features = c("countRFP", "disengagementScores", "entropyRFP", "floss", "fpkmRFP",
    "ioScore", "ORFScores", "RRS", "RSS", "startCodonCoverage", "startRegionCoverage",
    "startRegionRelative"),
  BPPARAM = bpparam()
)

Arguments

mainPath	folder for uORFome to put results
organism	scientific name of organism, like Homo sapiens, Danio rerio, etc.
df.rfp	ORFik experiment of Ribo-seq
df.rna	ORFik experiment of RNA-seq, set to NULL if you don't have RNA-seq
df.cage	ORFik experiment of CAGE, set to NULL if you don't have CAGE.
startCodons	default "ATG|CTG|TTG|GTG|AAG|AGG|ACG|ATC|ATA|ATT", set to "ATG|CTG|TTG|GTG" for a more certain set.
stopCodons	default "TAA|TAG|TGA"
mode	character, default: "uORF". alternative "aCDS". Do you want to predict on uORFs or artificial CDS. if "aCDS" will run twice once for whole length CDS and one for truncated CDS to validate model works for short ORFs. "CDS" is option to predict on whole CDS.
requiredActiveCds	numeric, default 30. How many CDSs are required to be detected active. Size of minimum positive training set. Will abort if not bigger than this number.
max.artificial.length	integer, default: 100, only applies if mode = "aCDS", so ignore this for most people, when creating artificial ORFs from CDS, how large should maximum ORFs be, this number is 1/6 of maximum size of ORFs (max size 600 if artificialLength is 100) Will sample random size from 6 to that number, if max.artificial.length is 2, you can get artificial ORFs of size (6, 9 or 12) (6, + 6 + (3x1), 6 + (3x2))
startCodons.cds.allowed	character, default same as startCodons argument. Which start codons can the CDS you train on have ?
stopCodons.cds.allowed	character, default same as stopCodons argument Which stop codons can the CDS you train on have ?
biomart	default "ensembl", get gene symbols and GO terms for uORF genes. Will be automaticly detected by organism name in ensembl database. Set to NULL if you don't want to check Gene symbols and GO terms.
features	features to train model on, any of the features created during ORFik::computeFeatures, default: c("countRFP", "disengagementScores", "entropyRFP", "floss", "fpkmRFP","ioScore", "ORFScores", "RRS", "RSS", "startCodonCoverage", "startRegionCoverage","startRegionRelative")
BPPARAM	An instance of a BiocParallelParam class, e.g., MulticoreParam, SnowParam, DoparParam.

Value

the prediction as data.table with 3 columns. Prediction (0 or 1), p0 (probability of a negtive prediction), p1 (probability of positive prediction). Only one of p0 and p1 can be > 0.5, and that value will decide if prediction is 0 or 1.

Examples

mainPath <- "~/bio/results/uORFome_Zebrafish"
# df.rfp <- read.experiment("path/to/rfp.csv")
# df.rna <- read.experiment("path/to/rna.csv") # Not required
# df.cage <- read.experiment("path/to/CAGE.csv") # Not required
# find_uORFome(mainPath, df.rfp, df.rna, df.cage)

Roleren/uORFomePipe documentation built on Jan. 14, 2024, 5:11 a.m.