I would question what you think you know about the presence of variables <=0.05 in your data.
How do you justify this knowledge ?
hits <- structure(list(`Combined adj. P-value` = c(0.04, 0.06), another_var = c(1,
2)), class = "data.frame", row.names = c(NA, -2L))
df.hits_sig1 <- hits[,hits$`Combined adj. P-value`<= 0.05]
# 0.04 0.06
df.hits_sig2 <- hits[hits$`Combined adj. P-value`<= 0.05,]
# Combined adj. P-value another_var
# 1 0.04 1
hits_no_p <- filter(hits,
`Combined adj. P-value`>0.05)
df.hits_sig3 <- hits_no_p[,hits_no_p$`Combined adj. P-value`<= 0.05]
#data frame with 0 columns and 1 row
df.hits_sig4 <- hits_no_p[hits_no_p$`Combined adj. P-value`<= 0.05,]
#[1] Combined adj. P-value another_var
#<0 rows> (or 0-length row.names)