Combining-Structure-Based-Features-and-Conserved-D.ppt

Combining Structure-Based Features and Conserved Data for MicroRNA Target Prediction by the Neural Network MethodShih-Yi Chao and Jung-Hsien Chiang*Department of Computer Science and Information Engineering National Cheng Kung University,Taiwan AgendanIntroductionnMaterialsnMethodsnExperimental ResultsnConclusionsIntroductionnWhat are microRNAs?short non-coding of about 1925 nucleotides RNAs single-stranded RNAs generated from endogenous transcripts that can form local hairpin structuresnWhat do the microRNAs do?regulating gene expression in animals and plants.taking part in post-transcriptional regulation either by arresting the translation of messenger RNAs(mRNAs)or by the cleavage of mRNAs.how they recognize and regulate target genes remains less well understood IntroductionnThe motivationit is difficult to identify the functions of short microRAN target sequences by only performing primary sequences alignment which means microRNAs contain only short complementary sequences,interrupted by gaps and mismatches.it may cause high false-positive rateMicroRNA target motifs are conserved more in structure than in primary sequences to consider the structure-based featuresmicroRNAs are often highly conserved across a wide range of species to consider the conserved data across speciesMaterialsnExtraction of 3UTR sequences in Human Data are originally downloaded from NCBI web site at.We extract 3UTRs sequences,gene ids,gene official symbol names,gene alias names,accession numbers,and so on The Human sequences are pairwisely aligned to all the C.elegans and Drrosophila target mRNA sequences and all sequences identified homology to C.elegans and Drrosophila are annotated as“potential homologue“and stored in our database.MaterialsnUsing the RNAfold program to predict the RNA secondary structures we use RNAfold to generate the structures predictions,which is the prediction of RNA secondary structure between the target sequence in the mRNA 3UTRs and the MicroRNAs.extracting the structure-based features such as bulges,interior loops and so on.Methods-Extracted Features IDThe descriptions of featuresDecoded value while inputting NN classifier1.AU match12GU match23GC match34mismatch45single nucleotide bulge56Non-single nucleotide bulge67gap78#of AU match at 5 part#9#of AU match at 3 part#10#of GU match at 5 part#11#of GU match at 3 part#12#of GC match at 5 part#13#of GC match at 3 part#14Total#of mismatch#15Total#of gap#16#of nucleotides within a non-single nucleotide bulge#System ProceduresExperimental ResultsGENE NAMEAccession NUMGO termsHomologyMicroRNAPARVBNM_013327protein bindingcytoskeleton cell adhesionC.elegansLet-7bGIPC1NM_005716receptor bindingprotein bindingmembrane fractionsoluble fractioncytosol G-protein coupledreceptor protein signaling pathway membraneC.elegansLet-7bFANCD2NM_001018115protein bindingNucleusChromosomeDNA repairCell cycleDrosophilaLet-7bTDO2NM_005651Function iron ion bindingtryptophan metabolismoxidoreductase activity neurotransmitter metabolismmetal ion bindingC.elegansLet-7b