discrepancy between R and awk fdr code snippets about protviz

[cpanse]

@ mariaderrico

cat ${COMET} \
  | awk -F"\t" 'NF==19{print $9"\t"$16}' \
  | cut -d"," -f1 \
  | sort -gr -k1 -S256M \
  | awk 'BEGIN{dc=0}$2~/^REV_/{dc++}{fdr=dc/(NR-dc);print dc"\t"NR"\t"fdr}'  \
  | awk '$NF<0.05{print}' \
  | tail

produces

251     5316    0.0495558
252     5317    0.0497532
252     5318    0.0497434
252     5319    0.0497336
253     5320    0.0499309
253     5321    0.0499211
253     5322    0.0499112
253     5323    0.0499014
253     5324    0.0498915
253     5325    0.0498817

and

library(protViz)
S <- lapply(F <- list.files(pattern="*.comet.txt"),
  function(f) read.table(f, header = TRUE, fill = TRUE, skip=1))  
rv <- do.call('rbind', lapply(S, protViz::summary.cometdecoy))

rv$mgf <- F  
knitr::kable(rv)

results in

nPSM	psmFdrCutoff	nDecoyPSM	nConfidentPSM	nDecoyPeptide	nConfidentPeptide	nDecoyProteins	nConfidentProteins	fdrPSM	fdrPeptide	fdrProtein	mgf
52158	0.05	254	5327	199	3352	199	1986	0.0476816	0.0593675	0.1002014	20201022_002_S223359_474-B4.comet.txt

5327 = 5325 + 2
Q.E.D.