This post is a stub. How can we predict the effects of alternative splicing (AS) events with the Ensembl Variant Effect Predictor (VEP)? In our recent paper in eLife, we did so by encoding them as if they were indels.

A unicellular relative of animals generates a layer of polarized cells by actomyosin-dependent cellularization (Dudin et al., eLife 2019)

First, load an ioe file generated with SUPPA2, that contains all AS events

ioe = read.table("input_file_from_suppa.ioe", header = T)
iot = ioe[ioe$seqname!="seqname",]
...

Encode AS events as indels and save as csv for later use in VEP:

psi = data.frame(
  seqname=iot$seqname,
  in_start=NA,
  in_end=NA,
  string=NA,
  in_strand=iot$ev_strand,
  event_id=iot$event_id
)


# SE: codified as deletion (XXX/-) that starts and ends where the exon should be
acceptable = iot$ev_type == "SE"
psi[acceptable,]$in_start  = iot[acceptable,]$ev_start
psi[acceptable,]$in_end    = iot[acceptable,]$ev_end
psi[acceptable,]$string    = paste(
  as.vector(subseq(x=    fi[psi[acceptable,]$seqname],
                   start=psi[acceptable,]$in_start,
                   end=  psi[acceptable,]$in_end)),
  "/-",
  sep=""
)
psi[acceptable,]$string    = paste(psi[acceptable,]$string,"/-",sep="")


# RI, A3, A5: codified as insertions (-/XXX) that start 
# at ev_start-1, and end at ev_start (reversed format!)
acceptable = 
  iot$ev_type == "RI" | 
  iot$ev_type == "A3" | 
  iot$ev_type == "A5"

psi[acceptable,]$in_start  = iot[acceptable,]$ev_start+2
psi[acceptable,]$in_end    = iot[acceptable,]$ev_start+1
psi[acceptable,]$string    = paste(
  "-/",
  as.vector(subseq(x =   fi[psi[acceptable,]$seqname],
                   start=iot[acceptable,]$ev_start,
                   end=  iot[acceptable,]$ev_end-1)),
  sep=""
)


# MX, AF, AL: codified as complex events (YY/XXX) that start 
# at ev_start-1, and end at ev_start (like insertions - reversed!)
acceptable = 
  iot$ev_type == "MX" | 
  iot$ev_type == "AF" | 
  iot$ev_type == "AL"

psi[acceptable,]$in_start  = iot[acceptable,]$ev_start+1
psi[acceptable,]$in_end    = iot[acceptable,]$ev_start
psi[acceptable,]$string    = paste(
  "-/",
  as.vector(subseq(x =   fi[psi[acceptable,]$seqname],
                   start=iot[acceptable,]$ev_start,
                   end=  iot[acceptable,]$ev_end)),
  "/",
  as.vector(subseq(x =   fi[psi[acceptable,]$seqname],
                   start=iot[acceptable,]$ev_startB,
                   end=  iot[acceptable,]$ev_endB)),
  sep=""
)


# format VEP pseudoindel file
psi = na.omit(psi)
write.table(psi,file="as_as_indels.csv", row.names = F,col.names = F, quote = F, sep = "\t")

Use the as_as_indels.csv file as input for VEP.

...

Bon profit!