Estimation of Gene Expression at Isoform-level from mRNA-Seq Data by Bayesian Hierarchical Modeling
mRNA-Seq is a precise and highly reproducible technique for measurement of transcripts levels and yields sequence information of a transcriptome at a single nucleotide base-level thus enabling us to determine splice junctions and alternative splicing events with high confidence. Often analysis of mR...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2012-11-01
|
| Series: | Frontiers in Genetics |
| Subjects: | |
| Online Access: | http://journal.frontiersin.org/Journal/10.3389/fgene.2012.00239/full |
| _version_ | 1798044047878651904 |
|---|---|
| author | Madhuchhanda eBhattacharjee Ravi eGupta Ramana eDavuluri |
| author_facet | Madhuchhanda eBhattacharjee Ravi eGupta Ramana eDavuluri |
| author_sort | Madhuchhanda eBhattacharjee |
| collection | DOAJ |
| description | mRNA-Seq is a precise and highly reproducible technique for measurement of transcripts levels and yields sequence information of a transcriptome at a single nucleotide base-level thus enabling us to determine splice junctions and alternative splicing events with high confidence. Often analysis of mRNA-Seq data does not attempt to quantify the expressions at isoform level. In this paper our objective would be use the mRNA-Seq data to infer expression at isoform level, where splicing patterns of a gene is assumed to be known. A Bayesian latent variable based modeling framework is proposed here, where the parameterization enables us to infer at various levels. For example, expression variability of an isoform across different conditions; the model parameterization also allows us to carry out two-sample comparisons, e.g. using a Bayesian t-test, in addition simple presence or absence of an isoform can also be estimated by the use of the latent variables present in the model. In this paper we would carry out inference on isoform expression under different normalization techniques, since it has been recently shown that one of the most prominent sources of variation in differential call using mRNA-Seq data is the normalization method used.The statistical framework is developed for multiple-isoforms and easily extends to reads mapping to multiple genes. This could be achieved by slight conceptual modifications in definitions of what we consider as a gene and what as an exon. Additionally proposed framework can be extended by appropriate modeling of the design matrix to infer about yet unknown novel transcripts. However such attempts should be made judiciously since the input date used in the proposed model does not use reads from splice junctions. |
| first_indexed | 2024-04-11T22:57:25Z |
| format | Article |
| id | doaj.art-7bd9348a2d9040d9926d684cb0fa4ded |
| institution | Directory Open Access Journal |
| issn | 1664-8021 |
| language | English |
| last_indexed | 2024-04-11T22:57:25Z |
| publishDate | 2012-11-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Genetics |
| spelling | doaj.art-7bd9348a2d9040d9926d684cb0fa4ded2022-12-22T03:58:19ZengFrontiers Media S.A.Frontiers in Genetics1664-80212012-11-01310.3389/fgene.2012.0023921011Estimation of Gene Expression at Isoform-level from mRNA-Seq Data by Bayesian Hierarchical ModelingMadhuchhanda eBhattacharjee0Ravi eGupta1Ramana eDavuluri2Pune UniversityThe Wistar InstituteThe Wistar InstitutemRNA-Seq is a precise and highly reproducible technique for measurement of transcripts levels and yields sequence information of a transcriptome at a single nucleotide base-level thus enabling us to determine splice junctions and alternative splicing events with high confidence. Often analysis of mRNA-Seq data does not attempt to quantify the expressions at isoform level. In this paper our objective would be use the mRNA-Seq data to infer expression at isoform level, where splicing patterns of a gene is assumed to be known. A Bayesian latent variable based modeling framework is proposed here, where the parameterization enables us to infer at various levels. For example, expression variability of an isoform across different conditions; the model parameterization also allows us to carry out two-sample comparisons, e.g. using a Bayesian t-test, in addition simple presence or absence of an isoform can also be estimated by the use of the latent variables present in the model. In this paper we would carry out inference on isoform expression under different normalization techniques, since it has been recently shown that one of the most prominent sources of variation in differential call using mRNA-Seq data is the normalization method used.The statistical framework is developed for multiple-isoforms and easily extends to reads mapping to multiple genes. This could be achieved by slight conceptual modifications in definitions of what we consider as a gene and what as an exon. Additionally proposed framework can be extended by appropriate modeling of the design matrix to infer about yet unknown novel transcripts. However such attempts should be made judiciously since the input date used in the proposed model does not use reads from splice junctions.http://journal.frontiersin.org/Journal/10.3389/fgene.2012.00239/fullBayesian latent variable modelingBayesian t-testisoform-expressionmRNA-seqmulti-sample comparisonspike-n-slab method |
| spellingShingle | Madhuchhanda eBhattacharjee Ravi eGupta Ramana eDavuluri Estimation of Gene Expression at Isoform-level from mRNA-Seq Data by Bayesian Hierarchical Modeling Frontiers in Genetics Bayesian latent variable modeling Bayesian t-test isoform-expression mRNA-seq multi-sample comparison spike-n-slab method |
| title | Estimation of Gene Expression at Isoform-level from mRNA-Seq Data by Bayesian Hierarchical Modeling |
| title_full | Estimation of Gene Expression at Isoform-level from mRNA-Seq Data by Bayesian Hierarchical Modeling |
| title_fullStr | Estimation of Gene Expression at Isoform-level from mRNA-Seq Data by Bayesian Hierarchical Modeling |
| title_full_unstemmed | Estimation of Gene Expression at Isoform-level from mRNA-Seq Data by Bayesian Hierarchical Modeling |
| title_short | Estimation of Gene Expression at Isoform-level from mRNA-Seq Data by Bayesian Hierarchical Modeling |
| title_sort | estimation of gene expression at isoform level from mrna seq data by bayesian hierarchical modeling |
| topic | Bayesian latent variable modeling Bayesian t-test isoform-expression mRNA-seq multi-sample comparison spike-n-slab method |
| url | http://journal.frontiersin.org/Journal/10.3389/fgene.2012.00239/full |
| work_keys_str_mv | AT madhuchhandaebhattacharjee estimationofgeneexpressionatisoformlevelfrommrnaseqdatabybayesianhierarchicalmodeling AT raviegupta estimationofgeneexpressionatisoformlevelfrommrnaseqdatabybayesianhierarchicalmodeling AT ramanaedavuluri estimationofgeneexpressionatisoformlevelfrommrnaseqdatabybayesianhierarchicalmodeling |