Context. A large fraction of stars host one or multiple close-in super-Earth planets. There is an active debate about whether these planets formed in situ or at greater distances from the central star and migrated to their current position. It has been shown that part of their observed properties (e.g., eccentricity distribution) can be reproduced by N-body simulations of in situ formation starting with a population of protoplanets of high masses and neglecting the effects of the disk gas. Aims. We plan to reassess the in situ formation of close-in super-Earths through more complete simulations. Methods. We performed N-body simulations of a population of small planetary embryos and planetesimals that include the effects of disk-planet interactions (e.g., eccentricity damping, Type I migration). In addition, we also consider the accretion of a primitive atmosphere from a protoplanetary disk. Results. We find that planetary embryos grow very quickly well before the gas dispersal, and thus undergo rapid inward migration, which means that one cannot neglect the effects of a gas disk when considering the in-situ formation of close-in super-Earths. Owing to their rapid inward migration, super-Earths reach a compact configuration near the disk's inner edge whose distribution of orbital parameters matches the observed close-in super-Earths population poorly. On the other hand, simulations including eccentricity damping, but no Type I migration, reproduce the observed distributions better. Including the accretion of an atmosphere does not help reproduce the bulk architecture of observations. Interestingly, we find that the massive embryos can migrate inside the disk edge while capturing only a moderately massive hydrogen/helium atmosphere. By this process they avoid becoming giant planets. Conclusions. The bulk of close-in super-Earths cannot form in situ, unless Type I migration is suppressed in the entire disk inside 1 AU.