Is predicting Drug-Target Interaction Possible When A Limited Amount of Bioactivity Data is Available?

Orta Doğu Teknik Üniversitesi, Enformatik Enstitüsü
Motivation Lately. deep learning-based models for drug-target interaction (DTI) prediction have yielded highly promising results. However. the majority of proposed deep learning models developed so far require large-scale training data. For many proteins. there is very little bioactivity data recorded in the databases or none at all; therefore. no prediction models are available for these. For example. in ChEMBL_29. only 8% of proteins have more than 1.000 bioactive compounds. while 56% of the proteins have less than 50 bioactive compounds. Deep transfer learning can be used to learn from such small amounts of data. Transfer learning is a machine learning approach where a model is trained for a source task. and this trained source model is then reused as an initial configuration to build (train) a target model for a different but related target task. Transfer learning has not been extensively exploited so far in the area of drug-target interaction prediction. In this study. we exploit the use of transfer learning techniques for the prediction of interactions between drugs/compounds and understudied target proteins that have scarce training data.


Serious game development methodology with system and human oriented approach
Çaplı, Levent Berke; Çakır, Serhat; Department of Science and Technology Policy Studies (2019)
Serious games are on the rise as a new and motivating education and training method. This rise is being supported by serious games’ ability to address individuals of all ages from many sectors. Although the game and education industries have established standards based on lessons learned, this is lacking for serious games. Considering the hybrid nature, it can be seen that there are many possible approaches for designing and developing serious games. It is possible to take elements form both simulation as w...
Explicit Characterizations for Plateaued-ness of p-ary (Vectorial) Functions
CARLET, CLAUDE; MESNAGER, SİHEM; Özbudak, Ferruh; SINAK, AHMET (2017-04-12)
Plateaued (vectorial) functions have an important role in the sequence and cryptography frameworks. Given their importance, they have not been studied in detail in general framework. Several researchers found recently results on their characterizations and introduced new tools to understand their structure and to design such functions. In this work, we mainly extend some of the observations made in characteristic 2 and given in (Carlet, IEEE Trans. Inf. Theor. 61(11), 6272- 6289, 2015) to arbitrary characte...
The impact of computer games on students' motivation
Üçgül, Memet; Çağıltay, Kürşat; Department of Computer Education and Instructional Technology (2006)
This study investigates the impact of computer games on students’ motivation. A computer game was created for this study by using Tomb Raider 4 Level Editor. The game covers photosynthesis, vitamins, carbohydrates, proteins and fat subjects of 5th grade science and technology course. Data was collected from 71 5th grade students of three primary schools. Students have played the game until the end of lesson than they were asked to complete The Instructional Materials Motivation Survey (IMMS). The study reve...
Feelings of language anxiety amongst non-native student teachers
Tum, Danyal Oztas (2012-02-11)
As rewarding as it may be. the teaching profession also carries with it considerable challenges and responsibilities. The transition from student teacher to teacher is a formidable step for student teachers. As student teachers approach the end of their teacher education programs and they become aware of the responsibilities and expectations of the profession, it is reasonable they feel overwhelmed, question their abilities, and experience feelings of inadequacy and anxiety. Such feelings can potentially ha...
Investigating the Sources of Collective Efficacy in Group Work of Preservice Science Teachers
Atasoy, Volkan; Çakıroğlu, Jale (2018-08-01)
Collective efficacy has been seen as one of the most powerful constructs among group-level motivational factors. The purpose of the present study was to examine effect of the source of collective efficacy on group behavior of preservice teachers. For this study, case study was conducted. Four preservice teachers who worked together in a one-semester science methods course were involved in the study. Semi-structured interviews were used as data collection. Open coding and axial coding was implemented for the...
Citation Formats
E. E. Orta Doğu Teknik Üniversitesi, “Is predicting Drug-Target Interaction Possible When A Limited Amount of Bioactivity Data is Available?,” Erdemli, Mersin, TÜRKİYE, 2022, p. 3063, Accessed: 00, 2023. [Online]. Available: