Analysis of Transcriptome Data and the Comparison of DNA Sequence - Assignment Example

BIOL5203M: The analysis of transcriptome data (two-channel arrays) and the comparison of DNA sequence Complete this pro forma in fewer than 200 words and submit by 4 pm on Thursday, 19th February 2008; a hard copy (with typed answers) to the Graduate School Office and an electronic copy to the VLE. Ancillary documents should be stapled to the hard-copy pro forma, and not sent as attachments with the electronic submission. Note that the Graduate School Office will not provide you with a stapler, and the complete pro forma should be in the form of Microsoft WORD. 1. Provide at least four explanations for why the average intensity of spots corresponding to one of the scans can be higher than the other without there being any biological variation. As with all answers, you should – where appropriate - provide references to support the material you provide. Expand space, as required. (5 marks) The variation in spot intensity can be caused by both systematic and stochastic errors (Le Crom, 2004). These errors come from the following: 1. Efficiency of RNA extraction and the reverse transcription process to produce the fluorophore-tagged cDNAs (targets) from mRNA. Differences in tissue and sensitivity to RNA degradation could affect the amount of targets formed. 2. Label effect – Cy3 and Cy5 are relatively unstable and could affect their respective incorporation during labelling which could result in differences in the spot intensities. 3. Dye-Label interaction – If the nucleotide sequence and secondary structure interact with the dye, the step could result in gene-label interactions that may be specific for some genes (for example, Cy3-dCTP could be preferentially incorporated compared to Cy5-dCTP). This will result in some genes that will always give more signal for one dye only (Tseng, Oh, Rohlin, Liao, & Wong, 2001). 4. Slide differences – Different slides could have different variations in imaging and hybridization. The amounts of probe DNA and labelled DNA could vary, so with the environment in the hybridization chamber . 5. Background noise on the slide and local surface curvature which could affect reading of the scanners (Tseng, Oh, Rohlin, Liao, & Wong, 2001). 6. Method of applying the spots (e.g. pin-spotted and inkjet). The method of applying the spots could lead to non-uniform spots and therefore variation. Pin-spotted arrays produce more variation compared to the inkjet method (Park, et al., 2004). 2. Normalise the data contained within the spreadsheet labelled BIOL5203M GEN. A couple of normalisation methods were described as part of the introductory lecture. Note that genes you suspect show changes in gene expression should not be used for normalisation, particularly when only a relatively small subset of genes is being analysed. Describe briefly the method you used to normalise the data. What genes, if any, did you remove for the purpose of normalisation? You can compare data to show the effects of excluding certain genes from the normalisation process. The plotting of trendlines is particularly informative in this example. Present the normalised data in a new column of the spreadsheet; make sure this column is clearly labelled. Credit will be given for normalisation procedures more sophisticated than making a correction based on the ratio between the average (or total) Cy3 and Cy5 signals. Attach any graphs, e.g. such as those used for regression analysis. Describe a simple experimental approach that could be used to assess non-biological sources of variation in two-channel arrays. (12 marks) a. The method used for normalization was the linear regression method. Outliers were sequentially removed to increase the R2 value, which is an indication of the goodness of fit of the linear regression where a perfectly linear line will have R2 = 1.0. b. The XY scatterplot for Cy5 and Cy3 was constructed. Figure 1 shows the plot for the original data set which shows a wide variation of spots. A trendline was also constructed (to provide the baseline information). Figure 1. Scatterplot for intensities of Cy3/Cy5 showing large variation of spots c. The outliers were removed, which, corresponded to the densities of following genes: folE, met (A,B,C,E,F,J,K), and the blank. The R2 value (measuring the goodness of fit off the data) was immensely improved from 0.193 to 0.952; showing that the linearity of the gene densities for Cy5 and Cy3 were normal. Figure 2. New scatterplot for the ratio of Cy3/Cy5 after removal of outlier data d. The Cy5 values were corrected based on the linear regression trendline where: [corrected Cy5 value = 1.535 * (uncorrected Cy5 value) – 198.4] e. The ratio for the expression mutant (Cy5) to wild-type (Cy3) was calculated based on the normalized Cy5 ( please refer to Table 1) f. Apart from the met genes, folE was the most overexpressed among all genes in the data set g. A simple experimental approach that could be used to assess the non-biological sources of variation in two-channel experiments: Generally there are two simple approaches used to determine non-biological and systematic sources of errors in a two-channel microarray experiment. One is to prepare enough RNA and cDNA samples which are prepared and labelled with both Cy5 and Cy3. Then this sample is divided equally and spotted in separate plates with several replications. Results will be able to show the presence of variations, which is a problem in most microarray experiments regardless of platform (Park, et al., 2004). The other approach is the dye-swap method where samples are alternately labelled; in one array, Cy3 is reference and Cy5 is sample; in another, the reverse is true. This is replicated 4-6 times to make sure that the cause of the technical variation is removed or identified. Table 1. Normalized values of gene densities based on Cy5 dye Gene Density Cy5 Normalized Density Cy5 (formula) Normalized Density Cy5 (values) Density Cy3 adhE 776 993 993 1010 argA 365 362 362 435 argB 400 416 416 505 argC 423 451 451 522 argD 581 694 694 752 argE 514 590 590 672 argF 691 862 862 928 argG 439 476 476 572 argH 369 368 368 473 argI 403 420 420 465 argR 308 274 274 426 argS 692 864 864 1130 argT 393 405 405 562 cpsA 383 389 389 233 crp 1423 1986 1986 2981 dnaG 357 350 350 353 fdhF 186 87 87 183 flhC 234 161 161 261 flhD 406 425 425 359 gatA 602 726 726 604 gatD 672 834 834 473 gatZ 3861 5729 5729 3844 glgA 516 594 594 632 glgC 346 333 333 445 glgP 334 314 314 434 glnA 400 416 416 655 glnH 680 846 846 1009 hfq 561 663 663 482 hycA 343 329 329 313 hypA 391 402 402 393 lacZ 2206 3188 3188 5420 metG 513 589 589 835 metH 384 392 392 387 metL 496 563 563 249 metR 739 936 936 612 miaA 712 894 894 627 mutL 327 303 303 297 nac 449 490 490 463 narU 247 181 181 277 ompA 3246 4784 4784 3611 pnp 776 993 993 813 polB 417 442 442 409 prlA 6853 10320 10320 9104 rnb 429 459 459 534 rne 386 393 393 359 rng 316 287 287 305 rpoA 9648 14611 14611 16080 rpoN 771 985 985 963 rpsO 707 886 886 807 rpSS 1402 1953 1953 2039 rspT 841 1092 1092 888 sspA 150 31 31 112 trpA 527 611 611 553 trpB 761 970 970 716 trpC 1070 1445 1445 1009 trpD 464 513 513 491 trpE 411 432 432 484 trpR 384 390 390 480 trpS 962 1278 1278 1789 trxA 638 782 782 711 uncB 1251 1721 1721 1439 uncE 826 1069 1069 1172 vacB 712 895 895 527 Sum (excluding the outliers 58070   76639 76671 folE 1262 371 metA 1372 429 metB 1450 649 metC 873 509 metE 22263 681 metF 2074 364 metJ 1646 4872 metK 3514 1002 Sum (including the outliers 92526     85549 Outliers density Cy5 density Cy3 Cy5/Cy3 blank 113 109 folE 1262 371 3.40682 metA 1372 429 3.19578 metB 1450 649 2.23314 metC 873 509 1.71517 metE 22263 681 32.68927 metF 2074 364 5.69281 metJ 1646 4872 0.33783 metK 3514 1002 3.50711 3. Calculate the ratio of expression for each gene and present in a new column, which should be clearly labelled. List the 10 genes with the largest increases in gene expression in the mutant. These genes can be easily identified by using the SORT function of Excel. Describe what the majority of these genes have in common; include a description of how at least some of these genes are known to be regulated at the level of transcription initiation. Include diagrams and referencing as deemed appropriate. (6 marks) a. The top ten genes with the highest Cy5/Cy3 ratios are the following: Order Genes Ratio Cy5/Cy3 1 metE 32.6893 2 metF 5.6928 3 metK 3.5071 4 folE 3.4068 5 metA 3.1958 6 metB 2.2331 7 metL 1.9879 8 metC 1.7152 9 cpsA 1.6401 10 gatD 1.4199 The top 8 genes are all involved in the biosynthetic pathway of methionine in E. coli (Figure 3). The met genes (E,F,K,A,B,L,C) are all part of the methionine regulon genes (reviewed in Marincs et al., 2006). Operators at the 5’ region of these genes are bound by the transcription repressor MetJ. metA encodes a homo-serine transsuccinylase converting L-homoserine to O-succinyl homoserine. metF gene encodes methylene tetrahydrofolate reductase that catalyses the formation of N5 methyltetrahydrofolate, a cofactor whic is involved in the conversion of homosysteine to methionine metK is an adenosyl transferase that converts methionine to the MetJ co-repressor, S-adenosyl methionine metB forms with metL, the only known operon in the E.coli met regulon MetC is involved in intermediate steps in the pathway metE had the highest change in expression level. MetE and MetH catalyze the conversion of homocysteine to methionine. FolE was also found to be upregulated ~4 times. FolE is involved in providing the cofactor for the final step in the methionine biosynthetic pathway. 4. Indicate what gene(s) when disrupted could cause the increases in gene expression that you have observed. Describe what you consider would be the simplest experimental approach that could be used to determine whether your choice was correct. (4 marks) Considering that majority of the genes that showed increased expression were involved in the methionine biosynthetic pathway, then the gene that was disrupted could be a transcription repressor. Studies have shown that this gene is metJ. MetJ, in the presence of its co-repressor AdoMet or S- adenosyl methionine, cooperatively binds operators in the 5’ region of at least 7 met genes comprising the met regulon, therefore repressing methionine synthesis. A very simple experimental approach to determine whether the choice of gene was correct is to perform a site-directed mutation on metJ. Site-directed mutagenesis, using the correct primers, directed to Q44K will alter one amino acid composition (based on literature) and hence the secondary structure of the MetJ protein. This mutation will change the protein cooperativity.The mutant gene will be cloned into plasmid and expressed in E. coli. Transcript levels of MetJ will determined (should be higher than wild-type) so with the transcript levels of the met genes in the regulon (this should be lower or very nil due to the repressor action). Comparison of gene expression can be performed by either simple real-time PCR or transcription profiling or microarray analysis. 5. To confirm that the gene for the transcription factor had been disrupted, a sequence reaction was primed using an oligonucleotide that annealed to the antibiotic-resistance gene in the cassette. A segment of the sequence obtained is provided at the bottom of this document. Perform a BLAST search at http://www.ncbi.nlm.nih.gov/BLAST/. Attach the first page of the list of hits from a BLAST search to your pro forma. Provide a brief description of the BLAST function and the statistical significant of the results you obtained. Describe what are meant by the terms homologue, paralogue and orthologue. What gene was identified as having the highest sequence similarity? Does this result meet with your expectations? If not, provide a possible explanation. Does the gene you identified have a homologue in E. coli? If you think it does, provide its name and a description of how you reached this conclusion. The bacterial species that was mutated is not indicated in the description to this practical. From the results of the BLAST alone, can you conclude it is definitely from Escherichia coli? (10 marks) BLAST results (please see attachment) The BLAST function (Basic Local Alignment Search Tool) is a service that seeks out from several databases genes that have sequence homology (Altschul, et al., 1997). Results for BLAST aresorted from the sequences with highest to the lowest homology. Indicators for degree of homology to the sequence submitted are the Query coverage and maximum identification, and the e-value (lower value means higher homology). Homologues is the term used for genes that have highly similar nucleotide sequences and biological function. Orthologues and paralogues are two types of homologues. Orthologues are genes in different species that are derived from a common ancestor. They may, or may not have the same function. Paralogues are homologous genes within a single species that have diverged by gene duplication. Thus, paralogues have similar bioligical functions. Paralogous genes may or may not have the same function. Based on the BLAST results, sequence similarities were highest for whole genomes. This is not expected because only a gene fragment was submitted for homology search. The nearest homologue was the E. coli MetJ gene coding for a regulatory protein (Accession No. M12869.1) The conclusion is reached based on the description of the gene given at Entrez Nucleotide (www.ncbi.nlm.nih.org) and its described fregulatory function which has been verified by the transcriptome microarray data. It cannot be concluded that the gene came from E.coli alone. The methionine biosynthetic pathway also occurs in other bacteria. It is highly probable that the gene could also have been isolated from Shigella because their values for all the statistical parameters in BLAST are the same. 6. Contrast and compare Affymetrix technology with the two-channel array system that was described during the lecture. (5 marks) Comparison of the Affymetrix technology with the two channel array system; 1. Both platforms generate data that is internally consistent, but the Affymetrix technology is better. 2. The two techniques do not have the same average values when comparing intensities for the same samples. 3. The two-channel microarray gives sharper contrast compared to Affymetrix. The Affymetrix technology will show more genes that are upregulated. 4. When mRNA is quantified by RT-PCR, there is a large difference in the values obtained in both arrays; although Affymetrix gives better correlation with RT-PCR. 5. There is no correlation between mRNA level and protein expression in both platforms. 6. The two-channel microarray is less expensive compared to Affymetrix; the DNA chips can contain high number of probes. However, since the data is for relative gene expression, normalization procedures are necessary. 7. The main advantage of the Affymetrix chip is that there is only one target that hybridizes with the chip and absolute expression is measured and compared to other genes. The Affymetrix technique does not require lengthy normalization. However, Affymetrix has to standardize the data analysis so that the data can be used to compare with other samples that are measured at a separate time. Sequence data: (Cut and paste into the Blast search) CGCACGCGTCGTCAGGTGAACAACCTGCGTCACGCTACCAACAGCGAGCTGCTGTGCGAAGCGTTTCTGCATGCCTTTACCGGGCAACCTTTGCCGGATGATGCCGATCTGCGTAAAGAGCGCAGCGACGAAATCCCGGAAGCGGCAAAAGAGATCATGCGTGAGATGGGGATTAACCCGGAGACGTGGGAATAC Do not forget to submit also your modified spreadsheet along with the hard copy of your work. Indicate the total number of works used to complete the assessment: _______________ References Altschul, SF, Madden, Y, Schaffer, A, Zhang, ., Zhang, J, Miller, W, and Lipman, D 1997, ‘Gapped BLAST and PSI-BLAST: A new generation of protein database search programs’, Nucleic Acids Research, vol. 25, pp. 3389-3402. Le Crom, S 2004, ‘Image Analysis and Normalization’ in Trascriptome Analyses: Experimental Design, Microarray Production and Data Analyses, FEBS and Ecole Normal Superiore: Paris Marincs, F, Manfield, I, Stead, J, Mcdowall, K, & Stockley, P 2006, ‘Transcript analysis reveals an extended regulon and the importance of protein–protein co-operativity for the Escherichia coli methionine repressor’, Biochemistry Journa , vol. 396, no. 2, pp. 227-234. Park, P, Caob, Y, Lee, S, Kim, J, Chang, M, Hart, R and Choi, S 2004, ‘Current issues for DNA microarrays: platform comparison,double linear amplification, and universal RNA reference’, Journal of Biotechnology, vol. 112, pp. 225-245. Tseng, G, Oh, M, Rohlin, L, Liao, J & Wong, W 2001, ‘Issues in cDNA microarray analysis: quality filtering, channel normalization,models of variation and assessment of gene effects’, Nucleic Acids Research, vol. 21, no.12, pp. 2549-2557. Weissbach, H & Brot, N 1991, ‘Regulation of methionine synthesis in Escherichia coli’, Molecular Microbiology, vol. 5, no. 7, pp. 1593-1597. Attachment : BLAST Results Bottom of Form Top of Form Accession Description Max score Total score Query coverage E value Max ident Links CU928163.2 Escherichia coli UMN026 chromosome, complete genome 352 352 100% 2e-94 100% CU928160.2 Escherichia coli IAI1 chromosome, complete genome 352 352 100% 2e-94 100% CU928145.2 Escherichia coli 55989 chromosome, complete genome 352 352 100% 2e-94 100% AP009240.1 Escherichia coli SE11 DNA, complete genome 352 352 100% 2e-94 100% CP001164.1 Escherichia coli O157:H7 str. EC4115, complete genome 352 352 100% 2e-94 100% CP001063.1 Shigella boydii CDC 3083-94, complete genome 352 352 100% 2e-94 100% CP000948.1 Escherichia coli str. K12 substr. DH10B, complete genome 352 352 100% 2e-94 100% CP000946.1 Escherichia coli ATCC 8739, complete genome 352 352 100% 2e-94 100% CP000800.1 Escherichia coli E24377A, complete genome 352 352 100% 2e-94 100% CP000802.1 Escherichia coli HS, complete genome 352 352 100% 2e-94 100% CP000266.1 Shigella flexneri 5 str. 8401, complete genome 352 352 100% 2e-94 100% AE005674.1 Shigella flexneri 2a str. 301, complete genome 352 352 100% 2e-94 100% AP009048.1 Escherichia coli str. K12 substr. W3110 DNA, complete genome 352 352 100% 2e-94 100% U00096.2 Escherichia coli str. K-12 substr. MG1655, complete genome 352 352 100% 2e-94 100% L19201.1 E. coli chromosomal region from 87.2 to 89.2 minutes 352 352 100% 2e-94 100% AE014073.1 Shigella flexneri 2a str. 2457T, complete genome 352 352 100% 2e-94 100% BA000007.2 Escherichia coli O157:H7 str. Sakai DNA, complete genome 352 352 100% 2e-94 100% AE005174.2 Escherichia coli O157:H7 EDL933, complete genome 352 352 100% 2e-94 100% CP000034.1 Shigella dysenteriae Sd197, complete genome 352 352 100% 2e-94 100% CP000036.1 Shigella boydii Sb227, complete genome 352 352 100% 2e-94 100% CP000038.1 Shigella sonnei Ss046, complete genome 352 352 100% 2e-94 100% M12869.1 E.coli metJ gene coding for a regulatory protein 352 352 100% 2e-94 100% M38202.1 E.coli regulon aporepressor mutant metJ193 allele, complete cds 352 352 100% 2e-94 100% CU928164.2 Escherichia coli IAI39 chromosome, complete genome 347 347 100% 1e-92 99% AF044502.1 Escherichia coli strain ec45 RhsF accessory genetic element core protein and dsORF-f3 genes, complete cds 347 347 98% 1e-92 100% CU928161.2 Escherichia coli S88 chromosome, complete genome 343 343 100% 1e-91 98% FM180568.1 Escherichia coli 0127:H6 E2348/69 complete genome, strain E2348/69 343 343 100% 1e-91 98% CP000970.1 Escherichia coli SMS-3-5, complete genome 343 343 100% 1e-91 98% CP000468.1 Escherichia coli APEC O1, complete genome 343 343 100% 1e-91 98% CP000243.1 Escherichia coli UTI89, complete genome 343 343 100% 1e-91 98% CU928158.2 Escherichia fergusonii ATCC 35469 chromosome, complete genome 338 338 100% 5e-90 98% CU928162.2 Escherichia coli ED1a chromosome, complete genome 334 334 100% 6e-89 97% CP000247.1 Escherichia coli 536, complete genome 334 334 100% 6e-89 97% AE014075.1 Escherichia coli CFT073, complete genome 334 334 100% 6e-89 97% CP001138.1 Salmonella enterica subsp. enterica serovar Agona str. SL483, complete genome 298 298 100% 5e-78 93% FM200053.1 Salmonella enterica subsp. enterica serovar Paratyphi A str. AKU_12601 complete genome, strain AKU_12601 298 298 100% 5e-78 93% CP001127.1 Salmonella enterica subsp. enterica serovar Schwarzengrund str. CVM19633, complete genome 298 298 100% 5e-78 93% CP001120.1 Salmonella enterica subsp. enterica serovar Heidelberg str. SL476, complete genome 298 298 100% 5e-78 93% Read More

Paralogues are homologous genes within a single species that have diverged by gene duplication. Thus, paralogues have similar bioligical functions. Paralogous genes may or may not have the same function. Based on the BLAST results, sequence similarities were highest for whole genomes. This is not expected because only a gene fragment was submitted for homology search. The nearest homologue was the E. coli MetJ gene coding for a regulatory protein (Accession No. M12869.1) The conclusion is reached based on the description of the gene given at Entrez Nucleotide (www.ncbi.nlm.nih.org) and its described fregulatory function which has been verified by the transcriptome microarray data.

It cannot be concluded that the gene came from E.coli alone. The methionine biosynthetic pathway also occurs in other bacteria. It is highly probable that the gene could also have been isolated from Shigella because their values for all the statistical parameters in BLAST are the same. 6. Contrast and compare Affymetrix technology with the two-channel array system that was described during the lecture. (5 marks) Comparison of the Affymetrix technology with the two channel array system; 1. Both platforms generate data that is internally consistent, but the Affymetrix technology is better. 2. The two techniques do not have the same average values when comparing intensities for the same samples. 3. The two-channel microarray gives sharper contrast compared to Affymetrix.

The Affymetrix technology will show more genes that are upregulated. 4. When mRNA is quantified by RT-PCR, there is a large difference in the values obtained in both arrays; although Affymetrix gives better correlation with RT-PCR. 5. There is no correlation between mRNA level and protein expression in both platforms. 6. The two-channel microarray is less expensive compared to Affymetrix; the DNA chips can contain high number of probes. However, since the data is for relative gene expression, normalization procedures are necessary. 7. The main advantage of the Affymetrix chip is that there is only one target that hybridizes with the chip and absolute expression is measured and compared to other genes.

The Affymetrix technique does not require lengthy normalization. However, Affymetrix has to standardize the data analysis so that the data can be used to compare with other samples that are measured at a separate time. Sequence data: (Cut and paste into the Blast search) CGCACGCGTCGTCAGGTGAACAACCTGCGTCACGCTACCAACAGCGAGCTGCTGTGCGAAGCGTTTCTGCATGCCTTTACCGGGCAACCTTTGCCGGATGATGCCGATCTGCGTAAAGAGCGCAGCGACGAAATCCCGGAAGCGGCAAAAGAGATCATGCGTGAGATGGGGATTAACCCGGAGACGTGGGAATAC Do not forget to submit also your modified spreadsheet along with the hard copy of your work.

Indicate the total number of works used to complete the assessment: _______________ References Altschul, SF, Madden, Y, Schaffer, A, Zhang, ., Zhang, J, Miller, W, and Lipman, D 1997, ‘Gapped BLAST and PSI-BLAST: A new generation of protein database search programs’, Nucleic Acids Research, vol. 25, pp. 3389-3402. Le Crom, S 2004, ‘Image Analysis and Normalization’ in Trascriptome Analyses: Experimental Design, Microarray Production and Data Analyses, FEBS and Ecole Normal Superiore: Paris Marincs, F, Manfield, I, Stead, J, Mcdowall, K, & Stockley, P 2006, ‘Transcript analysis reveals an extended regulon and the importance of protein–protein co-operativity for the Escherichia coli methionine repressor’, Biochemistry Journa , vol. 396, no. 2, pp. 227-234. Park, P, Caob, Y, Lee, S, Kim, J, Chang, M, Hart, R and Choi, S 2004, ‘Current issues for DNA microarrays: platform comparison,double linear amplification, and universal RNA reference’, Journal of Biotechnology, vol. 112, pp. 225-245.

Tseng, G, Oh, M, Rohlin, L, Liao, J & Wong, W 2001, ‘Issues in cDNA microarray analysis: quality filtering, channel normalization,models of variation and assessment of gene effects’, Nucleic Acids Research, vol. 21, no.12, pp. 2549-2557.

Read More