Bulanık c-ortalamalar kümeleme analizi ve sağlık alanında uygulaması
Yükleniyor...
Dosyalar
Tarih
2017
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Ege Üniversitesi, Sağlık Bilimleri Enstitüsü
Erişim Hakkı
info:eu-repo/semantics/openAccess
Özet
Kümeleme, benzer verileri gruplara ayırmak için kullanılan sınıflandırma yöntemidir. Kümeleme yöntemleri veri birimlerinin kümeye ait olma koşulları ya da kısıtlarına göre sert kümeleme ve yumuşak kümeleme yöntemleri olarak ikiye ayrılırlar. Yumuşak kümeleme yapan bulanık kümeleme algoritmaları ile veri birimleri sert kümelemede olduğu gibi yalnızca tek bir kümeye değil; farklı üyelik dereceleri ile açıklanan değerlere göre iki veya daha fazla kümeye girebilmektedir. Bu tez çalışmasında klasik kümeleme algoritmalarından farklı olarak, Bulanık C-ortalamalar kümeleme yöntemi mikroarray verilerine uygulanmıştır. Klasik yaklaşımlarda algoritmaların yapısı gereği yapılan kesin tanımlamalar bazı ilişkilerin gözden kaçmasına neden olabilmektedir. Bu nedenle gizli kalan ilişkilerin bulanık yöntemle ortaya konmasına çalışılmıştır.Mikro RNA'ların (miRNA, RNA: Ribo nükleik asit), çocuklarda akut lenfoblastik lösemi (ALL) üzerindeki etkisini incelemek üzere normal ve hasta dokulardan alınan mikrodizi tabanlı ifade verileri kullanılarak farklı sayıda miRNA kümelerinin Bulanık C-ortalamalar algoritmasıyla elde edilmesi hedeflenmiştir. Algoritmayla elde edilen farklı örneklem sayısına sahip, farklı kümelerdeki miRNA yapıları incelenmiştir. Bu yapılarda yer alan miRNA'larınmesajcı RNA (mRNA) yolaklarındaki ortak özellikleri araştırılmıştır. Bu çıktılar ile akut lenfoblastik lösemi ile ilişkili yolaklar bu algoritmaların çıktısını oluşturacak ve miRNA ile ilgili çalışmalarda yol gösterici olacaktır.
Clustering is a classification method to assign similar data elements into certain groups, called clusters. Clustering methods can be classified into two groups as "hard clustering" and "soft clustering" depending on the constraints on the memberhip degrees of data elements in clusters.Contrarily to hard clustering, in fuzzy clustering a data element is not only a member of unique cluster but also a member of the other clusters with varying degrees of membership.In this study, unlike classical clustering algorithms fuzzy C-means (FCM)clustering is applied to microarray data.Due to the classical approaches made according to the nature of the algorithms, exactly defined descriptions may cause some of the relationships to be overlooked.For this reason, hidden relations have been tried to be revealed via fuzzy method.The aim of this study was to evaluate the effect of miRNAs (micro RNA, RNA: Ribo nucleic aside) on children with ALL (Acute Lymphoblastic Leukemia) via microarray data following the use of fuzzy C-means of different number of miRNA clusters which was obtained from the analysis of the tissues in healthy and study cases. This study was made to evaluate the effect of miRNAs on children with all by using miRNA expression data on healthy and ill children (study cases) with sets containing different numbers of elements of fuzzy C-means algorithms. The miRNA structures in different clusters with different numbers ofsamples obtained via fuzzy clustering algorithmhave been evaluated.Structures of miRNAs in different sets and with different number of elements obtained from the algorithm were evaluated.The common characteristics of miRNAs of in each cluster in different messenger RNA (mRNA) pathways, have been investigated.miRNAs of this structure were investigated for their common properties on mRNAs pathways. In conclusion, new pathways associated with ALL may be described and those pathways may provide guidance to open up new horizons in the field of miRNA studies.
Clustering is a classification method to assign similar data elements into certain groups, called clusters. Clustering methods can be classified into two groups as "hard clustering" and "soft clustering" depending on the constraints on the memberhip degrees of data elements in clusters.Contrarily to hard clustering, in fuzzy clustering a data element is not only a member of unique cluster but also a member of the other clusters with varying degrees of membership.In this study, unlike classical clustering algorithms fuzzy C-means (FCM)clustering is applied to microarray data.Due to the classical approaches made according to the nature of the algorithms, exactly defined descriptions may cause some of the relationships to be overlooked.For this reason, hidden relations have been tried to be revealed via fuzzy method.The aim of this study was to evaluate the effect of miRNAs (micro RNA, RNA: Ribo nucleic aside) on children with ALL (Acute Lymphoblastic Leukemia) via microarray data following the use of fuzzy C-means of different number of miRNA clusters which was obtained from the analysis of the tissues in healthy and study cases. This study was made to evaluate the effect of miRNAs on children with all by using miRNA expression data on healthy and ill children (study cases) with sets containing different numbers of elements of fuzzy C-means algorithms. The miRNA structures in different clusters with different numbers ofsamples obtained via fuzzy clustering algorithmhave been evaluated.Structures of miRNAs in different sets and with different number of elements obtained from the algorithm were evaluated.The common characteristics of miRNAs of in each cluster in different messenger RNA (mRNA) pathways, have been investigated.miRNAs of this structure were investigated for their common properties on mRNAs pathways. In conclusion, new pathways associated with ALL may be described and those pathways may provide guidance to open up new horizons in the field of miRNA studies.
Açıklama
Anahtar Kelimeler
Bulanık C-Ortalamalar Yöntemi, Küme Geçerliliği, Kanser, Mirna Sorgulama Araçları, miRNA, Fuzzy C-Means Clustering, Cluster Validity, miRNA, Cancer, miRNA Target Prediction Tool
