One may think that scientific papers are easier to analyse than other categories of texts, since they are expected to talk about facts, hypotheses or methods in precise and explicit ways.
The following examples of sentences show that it is far from being the case. Sentences appear to be quite long and complex. A human reader might need to read them twice or more to grasp the main information they contain.
These sentences appeared in the comment fields of FlyBase database. They are part of a corpus of 1200 sentences which have been collected and analysed by Violaine Pillet during her PhD thesis.
Sentences which describe genetic interactions :
DNA binding studies in yeast systems suggest that the homeodomain is necessary for abd-A::Ecol\LexA proteins to bind to Ubx sites, but the homeodomain does not contact DNA exactly like bacterial helix-turn-helix proteins.
In-vitro transcription reactions and S1 nuclease analysis demonstrate the presence of z binding sites within 120bp region of twi promoter proximal to the transcription start site.
Results indicate that expression of ac stimulates expression of sc, and visa versa, therefore removal of one gene leads to the absence of both proneural gene products and sensory organs in the sites specificed by it cis-regulatory sequences.
Sentences which do not describe interactions:
Double mutants lacking the maternal activity of dorsalizing genes and zygotic dpp expression have a lateralized phenotype demonstrating that dpp activity is required early in development for the formation of dorsal epidermal tissue.
dpp imaginal discs were examined for evidence of cell death in mutants within the dpp disc region: loss of appendages is a result of massive apoptotic cell death in the corresponding regions of the imaginal disc during mid third larval instar.
Expression patterns were examined at the onset and completion of neurogenesis and results indicate that gsb-d expression is lineage specific suggesting that gsb-d functions as a selector gene in the patterning of the CNS.
Ectopic activity reflects inappropriate diffusion of the ligand to more central regions of the body, and therefore concluded that the tor receptor not only transduces the spatial signal imparted by the tor ligand, but sequesters the ligand, ensuring its correct localization.
The PRD-repeat domains of Nk1 and prd protein are sufficient to mediate protein-protein interaction, suggesting that the PRD-domain functions as a protein-binding interface and thereby may increase the DNA binding specificity of homeodomain transcription factors.