*3.10. Limitations*

The current study analyzed data available freely on-line and is based on the in silico approaches and analyses. The main disadvantage is the dependence on the quality of the experimental data from the external sources without own experimental validation. The overwhelming amount of biological data, stored and freely available for the research community worldwide; however, calls for in silico filtering of the content as first step. Experimental verification should be performed for the most relevant findings afterwards. We hope to experimentally verify some of the presented results in the future by ourselves or in cooperation with other research groups. Another limitation of the current work lies in the small number of genes studied. However, we aimed to focus on the clinically important and closely related genes of the ABCA subfamily where the available information is scarce. The other ABC gene subfamilies will be added in the ongoing study.

In the support of our gene family approach, it is important to mention that studies which deal with whole genome data sets require some compromises in the data mining. For example, they considered a uORF (or an uAUG) evolutionarily conserved when occurring in an orthologous transcript independently from its sequence, length, and position in the 5′UTR [32]; or, some analyses were performed only on transcripts containing a single uORF [29]. This, of course, simplifies the view on the genome complexity to some degree.
