Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “exon dynamics”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 325 records · Page 18Linked to original sources
PubMed2006

A mutation of spastin is responsible for swellings and impairment of transport in a region of axon characterized by changes in microtubule composition.

Mutations of the spastin gene (Sp) are responsible for the most frequent autosomal dominant form of spastic paraplegia, a disease characterized by the degeneration of corticospinal tracts. We show that a deletion in the mouse Sp gene, generating a premature stop codon, is responsible for progressive axonal degeneration, restricted to the central nervous system, leading to a late and mild motor defect. The degenerative process is characterized by focal axonal swellings, associated with abnormal accumulation of organelles and cytoskeletal components. In culture, mutant cortical neurons showed normal viability and neurite density. However, they develop neurite swellings associated with focal impairment of retrograde transport. These defects occur near the growth cone, in a region characterized by the transition between stable microtubules rich in detyrosinated alpha-tubulin and dynamic microtubules composed almost exclusively of tyrosinated alpha-tubulin. Here, we show that the Sp mutation has a major impact on neurite maintenance and transport both in vivo and in vitro. These results highlight the link between spastin and microtubule dynamics in axons, but not in other neuronal compartments. In addition, it is the first description of a human neurodegenerative disease which involves this specialized region of the axon.

Adenosine Triphosphatases↗
PubMed2000

Diverse splicing mechanisms fuse the evolutionarily conserved bicistronic MOCS1A and MOCS1B open reading frames.

Molybdenum is an essential cofactor in many enzymes, but must first be complexed by molybdopterin, whose synthesis requires four enzymatic activities. The first two enzymes of this pathway are encoded by the MOCS1 locus in humans. We describe here a remarkably well-conserved novel mRNA splicing phenomenon that produces both an apparently bicistronic MOCS1AM-OCS1B transcript, as well as a distinct class of monocistronic transcript. The latter are created by a variety of splicing mechanisms (alternative splice donors, alternative splice acceptors, and exon-skipping) to bypass the normal termination nonsense codon of MOCS1A resulting in fusion of the MOCS1A and MOCS1B open reading frames. Therefore, these "no-nonsense" transcripts encode a single bifunctional protein embodying both MOCS1A and MOCS1B activities. This coexpression profile was observed in vertebrates (human, mouse, cow, rabbit, opossum, and chicken) and invertebrates (fruit fly and nematode) spanning at least 700 million years of evolution. Our phylogenetic data also provide evidence that the bicistronic form of MOCS1 mRNA is likely to only produce MOCS1A protein and, combined with Northern analyses, suggests that MOCS1B is translated only as a fusion with MOCS1A. Taken together, the data presented here demonstrate a very highly conserved and physiologically relevant dynamic splicing scheme that profoundly influences the protein-coding potential of the MOCS1 locus.

Alternative Splicing↗
PubMed2005

A surfactant protein C precursor protein BRICHOS domain mutation causes endoplasmic reticulum stress, proteasome dysfunction, and caspase 3 activation.

BRICHOS is a domain found in several proteins consisting of approximately 100 amino acids with sequence and structural similarities. Mutations in BRICHOS domain have been associated with both degenerative and proliferative diseases in several nonpulmonary organs, although the pathogenic mechanisms are largely undefined. Recently, several mutations in surfactant protein C (SP-C) mapping to the BRICHOS domain located within the proprotein (proSP-C) have been linked to interstitial lung diseases. In vitro expression of one of these BRICHOS mutants, the exon 4 deletion (hSP-CDeltaexon4), promotes a dominant-negative perinuclear aggregation of the protein. The present study characterizes the trafficking behavior and pathogenic consequences resulting from hSP-CDeltaexon4 expression. Time-lapse and co-localization microscopy studies demonstrated enhanced green fluorescent protein (EGFP)/hSP-CDeltaexon4 expression in calnexin-positive (endoplasmic reticulum [ER]) compartment with subsequent time- and concentration-dependent development of ubiquitinated perinuclear inclusion bodies followed by apoptosis. Compared with controls, EGFP/hSP-CDeltaexon4 promoted upregulation of multiple ER stress species, activated caspase 3, and induced annexin V binding. Furthermore, in GFP-u cells, hSP-CDeltaexon4 directly inhibited proteasome activity. These results support a model whereby proSP-C BRICHOS mutations induce a dynamic toxic gain-of-function, causing apoptotic cell death both by early ER accumulation leading to an exaggerated unfolded protein response and by enhanced deposition of cellular aggregates associated with proteasome dysfunction.

Amino Acid Sequence↗
PubMed2003

The retinitis pigmentosa GTPase regulator (RPGR)- interacting protein: subserving RPGR function and participating in disk morphogenesis.

Retinitis pigmentosa is a photoreceptor degenerative disease leading to blindness in adulthood. Leber congenital amaurosis (LCA) describes a more severe condition with visual deficit in early childhood. Defects in the retinitis pigmentosa GTPase regulator (RPGR) and an RPGR-interacting protein (RPGRIP) are known causes of retinitis pigmentosa and LCA, respectively. Both proteins localize in the photoreceptor connecting cilium (CC), a thin bridge linking the cell body and the light-sensing outer segment. We show that RPGR is absent in the CC of photoreceptors lacking RPGRIP, but not vice versa. Mice lacking RPGRIP elaborate grossly oversized outer segment disks resembling a cytochalasin D-induced defect and have a more severe disease than mice lacking RPGR. Mice lacking both proteins are phenotypically indistinguishable from mice lacking RPGRIP alone. In vitro, RPGRIP forms homodimer and elongated filaments via interactions involving its coiled-coil and C-terminal domains. We conclude that RPGRIP is a stable polymer in the CC where it tethers RPGR and that RPGR depends on RPGRIP for subcellular localization and normal function. Our data suggest that RPGRIP is also required for disk morphogenesis, putatively by regulating actin cytoskeleton dynamics. The latter hypothesis may be consistent with a distant homology between the C-terminal domain of RPGRIP and an actin-fragmin kinase, predicted by fold recognition algorithms. A defect in RPGRIP encompasses loss of both functions, hence the more severe clinical manifestation as LCA.

Amino Acid Sequence↗
PubMed1997

Integrating database homology in a probabilistic gene structure model.

We present an improved stochastic model of genes in DNA, and describe a method for integrating database homology into the probabilistic framework. A generalized hidden Markov model (GHMM) describes the grammar of a legal parse of a DNA sequence. Probabilities are estimated for gene features by using dynamic programming to combine information from multiple sensors. We show how matches to homologous sequences from a database can be integrated into the probability estimation by interpreting the likelihood of a sequence in terms of the bit-cost to encode a sequence given a homology match. We also demonstrate how homology matches in protein databases can be exploited to help identify splice sites. Our experiments show significant improvements in the sensitivity and specificity of gene structure identification when these new features are added to our gene-finding system, Genie. Experimental results in tests using a standard set of annotated genes showed that Genie identified 95% of coding nucleotides correctly with a specificity of 91%, and 77% of exons were identified exactly.

Algorithms↗
PubMed2005

Structure, expression, and transcriptional regulation of the Strongylocentrotus franciscanus spec gene family encoding intracellular calcium-binding proteins.

The mechanisms by which gene expression patterns emerge during evolution are poorly understood. The sea urchin spec genes offer a useful means to investigate evolutionary mechanisms. Genes of the spec family from Strongylocentrotus purpuratus and Lytechinus pictus have identical patterns of aboral ectoderm-specific expression but exhibit species-specific differences in copy number, genomic structure, temporal expression, and cis-regulatory architecture. Here, we identify spec genes from a phylogenetic intermediate, Strongylocentrotus franciscanus, to gain insight into the evolution of the spec gene family and its transcriptional regulation. We identified two spec genes in the S. franciscanus genome, sfspec1a and sfspec1b, that were orthologous to spec1 from S. purpuratus. sfspec1b transcripts began to accumulate at the blastula stage and became progressively more abundant; this was reminiscent of spec expression in L. pictus but different from that in S. purpuratus. As expected, sfspec1b expression was restricted to aboral ectoderm cells. The six-exon structure of the sfspec1b genomic locus was identical to that of the S. purpuratus spec genes and was bounded by two repeat-spacer-repeat (RSR) repetitive sequence elements, which are conserved features of S. purpuratus spec genes and function as transcriptional enhancers. The enhancer activity of the sfspec1b RSRs was comparable to that of their S. purpuratus counterparts, although the placement and orientation of crucial cis-regulatory elements within the RSRs differed. We discovered a spec gene in S. franciscanus that was only distantly related to other spec genes but was highly conserved in S. purpuratus. Unexpectedly, this gene was expressed exclusively in endoderm lineages. Our results show that the evolution of spec cis-regulatory elements is highly dynamic and that substantial alterations can occur when maintaining or grossly modifying gene expression patterns.

Amino Acid Sequence↗
PubMed2005

Aggregation in Huntington's disease: insights through modelling.

Huntington's Disease (HD) is a late-onset, progressively degenerative brain disorder characterized by cell loss in the striatum and cortex. HD is caused by a polyglutamine (polyQ) expansion in the protein huntingtin (Htt). The mutant Htt is a substrate for caspases -2, -3 and -6. The cleavage of mutant Htt by caspase-2 has been suggested to underlie the selective neuronal death in HD. Once the mutant Htt is cleaved, a sticky and toxic fragment with the potential to form aggregates is released. The role of aggregation in the progression of HD has been extensively studied, yielding a plethora of ambivalent results. It has been shown that these are the diffuse, monomeric and oligomeric, forms of the mutant Htt fragment rather than the aggregates that are the major source of toxicity to the cells. We present here a mathematical model for aggregation in HD and discuss how it can relate to the selective neuronal death and the dependence of the disease onset on polyQ length. We describe the dynamical behavior of caspase-2, the release of monomeric forms of the mutant Htt fragment and the aggregation of these fragments through intermediate steps. Our model predicts that the concentration of toxic, intermediate oligomeric structures does not increase with increased caspase activity. We therefore suggest that the intermediate oligomeric forms of toxic Htt fragment do not account for selective and polyQ dependent neuronal death.

Age of Onset↗
PubMed2005

Inter- and intralocus recombination drive MHC class IIB gene diversification in a teleost, the three-spined stickleback Gasterosteus aculeatus.

The mutational mechanism underlying the striking diversity in MHC (major histocompatibility complex) genes in vertebrates is still controversial. In order to evaluate the role of inter- and intragenic recombination in MHC gene diversification, we examined patterns of nucleotide polymorphism across an exon/intron boundary in a sample of 31 MHC class IIB sequences of three-spined stickleback (Gasterosteus aculeatus). MHC class IIB genes of G. aculeatus were previously shown to be under diversifying (positive) selection in mate choice and pathogen selection experiments. Based on recoding of alignment gaps, complete intron 2 sequences were grouped into three clusters using maximum-parsimony analysis. Two of these groups had >90% bootstrap support and were tentatively assigned single locus status. Intron nucleotide diversity within and among loci was low (p-distance within and among groups = 0.016 and 0.019, respectively) and fourfold lower than the rate of silent mutations in exon 2, suggesting that noncoding regions are homogenized by frequent interlocus recombination. A substitution analysis using GENECONV: revealed as many intergenic conversion events as intragenic ones. Recombination between loci may explain the occurrence of sequence variants that are particularly divergent, as is the case in three-spined stickleback, with nucleotide diversity attaining dN = 0.39 (peptide-binding residues only). For both MHC class II loci we also estimated the amount of intragenic recombination as population rate (4N(e)r) under the coalescent and found it to be approximately three times higher compared to point mutations (Watterson estimate per gene, 4N(e)mu). Nonindependence of molecular evolution across loci and frequent recombination suggest that MHC class II genes of bony fish may follow different evolutionary dynamics than those of mammals. Our finding of widespread recombination suggests that phylogenies of MHC genes should not be based on coding segments but rather on noncoding introns.

Animals↗
PubMed1988

Structure of a pre-mRNA branch point/3' splice site region.

We have analyzed the solution structure of RNA containing the branch point/3' splice site region of the first intervening sequence (IVS1) of human beta-globin pre-mRNA by nuclease mapping and NMR. Nuclease mapping indicates that there are two distinct structural domains: one contains the branch point region, and the other includes the 3' splice site and second exon sequences. In the branch point domain, the adenosine at which the RNA branch forms appears to be in the loop of a stem/loop structure. The branch point structure does not appear to interact with other parts of the RNA, since its unique nuclease digestion pattern is conserved among transcripts containing the entire intron or only the branch point region. This is confirmed by a comparison of the NMR spectra of two RNA transcripts; a distinct set of resonances appears in the spectra of the RNA containing only branch sequences or including 3' splice site/exon 2 sequences. NMR studies further show that the 3' splice site/exon 2 domain has a lower melting temperature than the branch point domain, suggesting that the two regions are distinct dynamically as well as structurally. Nuclease mapping studies of adenovirus major late IVS1 indicate that this RNA has structural features in common with the human beta-globin transcript.

Base Sequence↗
PubMed2003

Founder von Willebrand factor haplotype associated with type 1 von Willebrand disease.

To date, no dominant mutation has been identified in a significant proportion of patients with type 1 von Willebrand disease (VWD). In this study, we examined 70 families as part of the Canadian Type 1 VWD Study. The entire VWF gene was sequenced for 1 index case, revealing 2 sequence variations: intron 30 (5312-19A>C) and exon 28 at Tyr1584Cys (4751A>G). The Tyr1584Cys variation was identified in 14.3% (10 of 70) of the families and was in phase with the 5312-19A>C variation in 7 (10.0%) families. Both variants were observed in 2 of 10 UK families with type 1 VWD, but neither variant was found in 200 and 100 healthy, unrelated persons, respectively. Mean von Willebrand factor antigen (VWF:Ag), VWF ristocetin cofactor (VWF:RCo), and factor VIII coagulant activity (FVIII:C) for the index cases in these families are 0.4 U/mL, 0.36 U/mL, and 0.54 U/mL, respectively, and VWF multimer patterns show no qualitative abnormalities. Aberrant VWF splicing was not observed in these patients, and both alleles of the VWF gene are expressed as RNA. Molecular dynamic simulation was performed on a homology model of the VWF-A2 domain containing the Tyr1584Cys mutation. This showed that no significant structural changes occur as a result of the substitution but that a new solvent-exposed reactive thiol group is apparent. Expression studies revealed that the Tyr1584Cys mutation results in increased intracellular retention of the VWF protein. We demonstrate that all the families with the Tyr1584Cys mutation share a common, evolved VWF haplotype, suggesting that this mutation is ancient. This is the first report of a mutation that segregates in a significant proportion of patients with type 1 VWD.

Alleles↗
PubMed2002

Structure and expression of three Emx genes in the dogfish Scyliorhinus canicula: functional and evolutionary implications.

We report the characterization of three Emx genes in a chondrichthyan, the dogfish Scyliorhinus canicula. Comparisons of these genes with their osteichthyan counterparts indicate that the gnathostome Emx genes belong to three distinct orthology classes, each containing one of the dogfish genes and either the tetrapod Emx1 genes (Emx1 class), the osteichthyan Emx2 genes (Emx2 class) or the zebrafish Emx1 gene (Emx3 class). While the three classes could be retrieved from the pufferfish genome data, no indication of an Emx3-related gene in tetrapods could be found in the databases, suggesting that this class may have been lost in this taxon. Expression pattern comparisons of the three dogfish Emx genes and their osteichthyan counterparts indicate that not only telencephalic, but also diencephalic Emx expression territories are highly conserved among gnathostomes. In particular, all gnathostomes share an early, dynamic phase of Emx expression, spanning presumptive dorsal diencephalic territories, which involves Emx3 in the dogfish, but another orthology class, Emx2, in tetrapods. In addition, the dogfish Emx2 gene shows a highly specific expression domain in the cephalic paraxial mesoderm from the end of gastrulation and throughout neurulation, which suggests a role in the segmentation of the cephalic mesoderm.

Amino Acid Sequence↗
PubMed2026

Genomic distribution characteristics and interspecific differences of microsatellite landscapes in Felidae.

BACKGROUND: Microsatellites within genomes play crucial roles in regulating gene expression, DNA replication, and chromosomal structure and function. Analyzing the composition and distribution patterns of microsatellites in closely related species not only reveals their evolutionary dynamics and adaptive mechanisms but also provides essential technical support for applications in genetic breeding, species conservation, and disease research. As one of the world's most captivating animal groups, the landscape patterns of microsatellites across feline genomes remain to be systematically characterized. RESULTS: This study utilized high-quality genomic data to conduct a systematic comparative analysis of microsatellite landscape distribution patterns across the genomes of 13 felid species. The findings revealed that microsatellite abundance and distribution exhibit species-specific characteristics, with a non-random genomic distribution and a negative correlation between microsatellite abundance and repeat length. The predominant distribution pattern followed the sequence: single > double > quadruple > triple > quintuple > sextuple nucleotide repeats. Microsatellite abundance peaked in intergenic regions, whereas trinucleotide repeats were more prevalent within exons. Coding regions showed a marked preference for trinucleotide and hexanucleotide repeats. Enrichment analysis of GO and KEGG pathways indicated that coding sequences containing microsatellites were primarily involved in transcription and translation processes. CONCLUSIONS: Our study elucidates the distribution patterns and characteristics of microsatellites across diverse feline species, providing significant insights into their evolutionary mechanisms and functional roles. Furthermore, these findings establish a valuable reference and foundational dataset for the future development of high-quality, species-specific microsatellite markers in felids.

Animals↗
PubMed2005

Fixation of conserved sequences shapes human intron size and influences transposon-insertion dynamics.

The basis for intron expansion in humans is largely unexplored. In this article, we demonstrate that intron expansion has primarily been determined by fixation of multispecies conserved sequences (MCSs) over time. The presence of MCSs has shaped intron features: the insertion of transposable elements (TEs) has been constrained as more MCSs were fixed. Analysis of TE and MCS distribution suggested an unprecedented estimate of information requirements for proper splicing of long introns with indication of sequence constraints extending up to >3 kb downstream 5' splice sites.

Alternative Splicing↗
PubMed2026

Genome-wide variation analysis of two Salvia hispanica L. genotypes and implication for associations with metabolic and adaptive traits.

BACKGROUND: Advances in next-generation sequencing have accelerated genome-wide exploration of genetic diversity in underutilized oilseed crops. Salvia hispanica L. (chia), a high-nutrient pseudocereal rich in omega-3 fatty acids, is increasingly valued for its health benefits and commercial potential, yet it remains poorly characterized at the genomic level. Understanding the scale and nature of genomic variation is essential for improving complex traits such as oil yield, stress tolerance, and seed quality. METHODS: Two contrasting chia genotypes, Black-chia (CACH-B) and White- chia (CACH-W), were resequenced using the Bio-Resequencing Toolkit (BRT) pipeline. High-coverage sequencing, with a mapping rate exceeding 99% and an average depth of approximately 28×, facilitated the detection and annotation of single-nucleotide polymorphisms (SNPs), insertions and deletions (InDels), copy-number variations (CNVs), and structural variants (SVs). The functional classification of variant impacts enabled the identification of genes potentially linked to metabolic and adaptive traits. RESULTS: A total of 1.97 million SNPs, 401,493 InDels, 836 CNVs, and 15,288 SVs were identified across the chia genome. Notably, approximately 53% of exonic SNPs were non-synonymous (dN/dS ≈ 1.28), predominantly affecting lipid metabolism, transcriptional regulation, and stress response pathways, potentially altering key agronomic traits. In addition, CNV hotspots were concentrated in chromosomes 3 and 6, overlapping MYB, WRKY, and bZIP transcription factor loci, may potentially be involved in stress tolerance and yield. Furthermore, structural rearrangements, including inversions and duplications within the FAD2, FAD3, and CYP450 gene clusters, were potentially associated with seed pigmentation and omega-3 biosynthesis, pointing to their potential breeding relevance. Observed heterozygosity (Hₒ ≈ 0.71) and nucleotide diversity (π ≈ 7 × 10-3) indicated moderate to high allelic richness. In addition, the low FST value (0.038) indicates substantial genomic similarity between the two genotypes. CONCLUSION: This study presents the first comprehensive map integrating SNPs, CNVs, and SVs in S. hispanica L. The results reveal a structurally dynamic genome characterized by substantial sequence and structural variation, providing valuable insights into genomic diversity and potential adaptive mechanisms in chia. The coexistence of high SNP diversity and abundant structural variation underpins chia's nutritional specialization and environmental resilience. These results deliver a foundational genomic resource for marker-assisted breeding, genome-wide association studies, and the development of climate-resilient chia cultivars.

Copy-number variation, structural variation↗
PubMed2001

Hes7: a bHLH-type repressor gene regulated by Notch and expressed in the presomitic mesoderm.

BACKGROUND: Whereas Notch signalling is essential for somitogenesis, mice deficient for the basic helix-loop-helix (bHLH) genes Hes1 and Hes5, downstream Notch effectors, display normal somite formation, indicating that there may be an as-yet unidentified Hes1-related bHLH gene. RESULTS: We identified a novel bHLH gene, designated Hes7, from mouse embryos. Hes7 has a conserved bHLH domain in the amino-terminal region and the WRPW domain at the carboxy-terminal end, like Hes1. The mouse Hes7 gene is located next to Aloxe3, which is mapped to a position 37.0 cM from the centromere on chromosome 11. In a transfection analysis, Hes7 represses transcription from the N box- and E box-containing promoters. In addition, Hes7 suppresses the E47-induced transcriptional activation. Promoter analysis indicated that Hes7 expression is controlled by Notch signalling. Strikingly, Hes7 is specifically expressed in the presomitic mesoderm in a dynamic manner. We also identified two related bHLH genes from human: one is closely related to mouse Hes7 and therefore designated hHes7 and the other designated hHes4. CONCLUSION: The structure, transcriptional activity and expression pattern in the presomitic mesoderm of Hes7 are very similar to those of Hes1, suggesting that Hes7, together with Hes1, may play a role in somite formation under the control of Notch signalling.

Amino Acid Sequence↗
PubMed2004

eShadow: a tool for comparing closely related sequences.

Primate sequence comparisons are difficult to interpret due to the high degree of sequence similarity shared between such closely related species. Recently, a novel method, phylogenetic shadowing, has been pioneered for predicting functional elements in the human genome through the analysis of multiple primate sequence alignments. We have expanded this theoretical approach to create a computational tool, eShadow, for the identification of elements under selective pressure in multiple sequence alignments of closely related genomes, such as in comparisons of human-to-primate or mouse-to-rat DNA. This tool integrates two different statistical methods and allows for the dynamic visualization of the resulting conservation profile. eShadow also includes a versatile optimization module capable of training the underlying Hidden Markov Model to differentially predict functional sequences. This module grants the tool high flexibility in the analysis of multiple sequence alignments and in comparing sequences with different divergence rates. Here, we describe the eShadow comparative tool and its potential uses for analyzing both multiple nucleotide and protein alignments to predict putative functional elements.

Animals↗
PubMed2002

Influence of an asparagine to lysine mutation at amino acid 3516 of apolipoprotein B on low-density lipoprotein receptor binding.

BACKGROUND: Three mutations in the apolipoprotein B (apoB) gene have previously been established as important causes of impaired receptor binding of LDL and, hence, Familial Defective Apolipoprotein B 100 (FDB). Previously, undescribed mutations were sought. METHODS: Using denaturing gradient gel electrophoresis for mutation detection, DNA from 1852 new patients was examined. RESULTS: A previously undiscovered mutation was found in codon 3516, located between known FDB mutations at codons 3500 and 3531. The new mutation introduces a positively charged amino acid-lysine-while other FDB mutations remove a positively charged residue, arginine. The phenotype was intriguing, LDL derived from N3516K heterozygotes allowed only poor growth of an LDL cholesterol-dependent cell line. ApoB-100-specific antibody MB47 bound to LDL from N3516K heterozygotes with increased affinity indicating a probable conformational change caused by the substitution. In contrast to these results, a competitive displacement assay in fibroblasts showed normal (or better) binding affinity to LDL receptors and using dynamic laser scattering no preferential accumulation of 3516K LDL particles in plasma was found. CONCLUSION: Discovery of the mutation and characterisation of N3516K LDL reveals another naturally occurring apoB mutation that influences conformation of LDL apoB and its interaction with the LDL receptor.

Adult↗
PubMed2003

HMM sampling and applications to gene finding and alternative splicing.

The standard method of applying hidden Markov models to biological problems is to find a Viterbi (maximal weight) path through the HMM graph. The Viterbi algorithm reduces the problem of finding the most likely hidden state sequence that explains given observations, to a dynamic programming problem for corresponding directed acyclic graphs. For example, in the gene finding application, the HMM is used to find the most likely underlying gene structure given a DNA sequence. In this note we discuss the applications of sampling methods for HMMs. The standard sampling algorithm for HMMs is a variant of the common forward-backward and backtrack algorithms, and has already been applied in the context of Gibbs sampling methods. Nevetheless, the practice of sampling state paths from HMMs does not seem to have been widely adopted, and important applications have been overlooked. We show how sampling can be used for finding alternative splicings for genes, including alternative splicings that are conserved between genes from related organisms. We also show how sampling from the posterior distribution is a natural way to compute probabilities for predicted exons and gene structures being correct under the assumed model. Finally, we describe a new memory efficient sampling algorithm for certain classes of HMMs which provides a practical sampling alternative to the Hirschberg algorithm for optimal alignment. The ideas presented have applications not only to gene finding and HMMs but more generally to stochastic context free grammars and RNA structure prediction.

Algorithms↗