Galaxy formation and evolution has been intensively studied in the past two decades. Resultsand discoveries revolutionized the field and we now have much better understandingof how these huge conglomerates of baryonic and dark matter evolve.In the framework of two different projects, which constitute the backbone of thisthesis, I investigated the nature of almost 100 emission-line galaxies selected with thenarrow-band selection method. These candidates can be: galaxies with Hα emissionlines at redshift z ∼ 0.6, [Oiii]/Hβ emission-line galaxies at redshift z ∼ 1.15 and [Oii]emitters at z ∼ 1.85.I used the near-infrared data from NB1060, Y and J filters to perform colour-colourand colour-magnitude selections. With the broad-band data from CANDELS catalogue,I performed SED fitting and derived photometric redshifts and other physical propertiesfor the candidate emission-line galaxies. Significant differences between the two selectionmethods have been found. The colour-colour selection method, tends to pick galaxieswith high colour excess and can leave some strong emission-line candidates with relativelylower colour excess out of the sample. The populations of selected galaxies can also bevery biased. The colour-magnitude selection method picks not only “normal” galaxies,but also starburst ones.I investigated the physical properties and colour indices for the selected galaxies,obtaining the following results. I) Stellar masses for narrow-band selected galaxies arefound to be in the range M∗ ∼ 107 − 1011M⊙. The derived star-formation rates suggestthat they can be either normal galaxies in the main sequence of the M∗ vs. SFR relation,or starburst ones. II) A steepening of the M∗ vs. SFR slope for the galaxies in thelower part of the mass range (M∗ < 109.4M⊙) is found. III) From studying the spatialdistribution of our sample the discovery of a galaxy protocluster at the redshift of z = 1.85and the likely discovery of a filament at z = 1.15 are reported. The identification ofthese gravitationally bound structures holds regardless of selection method employed inthe analysis.In this thesis, I also briefly summarise the first results of an on-going investigation ofgalaxies selected according to their robust spectroscopic redshifts from the large MUSYCmultiband survey. The aim of this project is to study the evolution of the physicalproperties of galaxies across large redshift range (0 < z < 6) and see their implicationson the evolution of the universe.v