Comparison of Bayesian Hierarchical Mixture Model With Structural Mixed Model for Determination Differentially Expressed Genes in Terms of Feed Efficiency in Rainbow Trout
Abstract
Bu çalışma, beslenme etkinliği bakımından iyi ve kötü performans gösteren iki klon hattı genotipten elde edilen gen ifade seviyelerinin yapısal karışık ve bayesci karışımlı modellerin kullanımıyla belirlenmesi ve modeller arasında karşılaştırma yapılması amacıyla gerçekleştirilmiştir. Araştırmada kullanılan alabalıklar Gournay-sur-Aronde INRA(Fransa Ulusal Tarım Araştırma Enstitüsü)'da yetiştirilen 10 genetik klon genotipinden en iyi ve en kötü beslenme etkinliği performansına sahip iki klon hattı seçilerek elde edilmiştir. İki klon hattının belirlenmesinde tanımlayıcı istatistikeler, varyans analiz yöntemleri ve karışık model eşitliklerinden yararlanılmıştır.Seçimi yapılan iki genotipten karaciğer dokularından örnekler alınarak mRNA ektraksiyonu yapılmış ve naylon en küçük dizilimler kullanılarak gen ifadeleri çalışmadaki her bir birey için elde edilmiştir. En iyi beslenme etkinliğine sahip genotip için 10, en kötü beslenme etkinliğine sahip genotip için 7 birey kullanılmıştır. Elde edilen gen ifade değerlerinin normalizasyonu için yine karışık model eşitlikleri kullanılmıştır.Yapısal karşık model ile %5 FDR oranında 9, %10 FDR oranında ise 20 genin farklı olarak ifade edildiği saptanmıştır. Bayesci karışımlı modelde ise farklı parametrik dağılımlar kullanılarak 6 farklı model oluşturulmuştur. Bu altı modelin DIC, sonsal posterior ortalamalar ve etkili parametre sayısı gibi istatistikler dikkate alınarak en iyi uyuma sahip bayes karışımlı model belirlenmiştir. Seçilen model farklı ifade edilen ve edilmeyen genler için Gamma önsel dağılımını kullanırken varyanslar için log Normal dağılımı kullanmaktadır. Model 2 olarak belirlenen bayes karışımlı model ile % 5 FDR düzeyinde 16, %10 FDR düzeyinde ise 38 genin farklı olarak ifade edildiği belirlenmiştir.Her iki model ile elde edilen % 10 FDR düzeyindeki genler karşılaştırıldığında 6 genin her iki modelde de önemli bulunduğu belirlenmiştir.
This study was carried out to make comparison between structural mixed model and Bayesian mixture model and to determine the gene expression performance that obtained two clone line genotype in terms of feed efficiency in rainbow trout fish population. The trouts were used in the research obtained by selection from 10 clonal genotype that raised in Gournay-sur-Aronde- INRA. To determine two clone lines, various analytical procedures and models including descriptive statistics, analysis of variance, and mixed model equations were utilized..mRNA was extracted using samples of liver tissue that came from two selected genotypes and then gene expression values was obtained using nylon membranes for each individual in study. In the microarray data analyses part, top 10 and worst 7 individuals performed for feed efficiency were used. . Mixed model equation was used for normalization of gene expression measurements.After fitting the structural mixed model, 20 and 9 differentially expressed genes were determined at level of 10 % and 5 % FDR, respectively. In Bayesian mixture model, 6 different models were tried for different parametric distributions. Best fitted model was determined by using DIC, posterior means deviance, and effective number of parameters criteria. The gamma prior was used for elected model for different and non-different expressed genes and a log Normal prior was assumed for gene variances. The selected model which called Model 2 determined that 38 and 16 genes expressed differentially at 10 % and 5 % FDR levels, respectively.Six genes, among differentially expressed genes, were found to be important in the both models at %10 FDR level.
This study was carried out to make comparison between structural mixed model and Bayesian mixture model and to determine the gene expression performance that obtained two clone line genotype in terms of feed efficiency in rainbow trout fish population. The trouts were used in the research obtained by selection from 10 clonal genotype that raised in Gournay-sur-Aronde- INRA. To determine two clone lines, various analytical procedures and models including descriptive statistics, analysis of variance, and mixed model equations were utilized..mRNA was extracted using samples of liver tissue that came from two selected genotypes and then gene expression values was obtained using nylon membranes for each individual in study. In the microarray data analyses part, top 10 and worst 7 individuals performed for feed efficiency were used. . Mixed model equation was used for normalization of gene expression measurements.After fitting the structural mixed model, 20 and 9 differentially expressed genes were determined at level of 10 % and 5 % FDR, respectively. In Bayesian mixture model, 6 different models were tried for different parametric distributions. Best fitted model was determined by using DIC, posterior means deviance, and effective number of parameters criteria. The gamma prior was used for elected model for different and non-different expressed genes and a log Normal prior was assumed for gene variances. The selected model which called Model 2 determined that 38 and 16 genes expressed differentially at 10 % and 5 % FDR levels, respectively.Six genes, among differentially expressed genes, were found to be important in the both models at %10 FDR level.
Description
Keywords
Biyoistatistik, Genetik, Istatistik, Bayes Analizi, Normalleştirme, Biostatistics, Genetics, Statistics, Bayes Analysis, Normalization
Turkish CoHE Thesis Center URL
WoS Q
Scopus Q
Source
Volume
Issue
Start Page
End Page
84