Examples
This page serves as an introduction to the use of AYM on your dataset or model.
We present the usages and outcomes of the annotation and retrieval processes for different omics datasets.
In case you may encounter time out problems with the model retrieval or the local alignments, try checking the "Only search-relevant annotations" box in the "Annotate automatically" menu.
This will force the annotation routine to use only those annotations which are related to annotations in the BioModels database. A retrieval after using this option will lead to a faster
processing, but it might also change the results slightly and make them more significant.
Data set Klevecz et al (introduction) (click to expand) | Publication title | A genomewide oscillation in transcription gates DNA replication and cell cycle |
---|
Authors | Klevecz, Bolen, Forrest, and Murray |
---|
Supplementary material | Supporting table 1 |
---|
Results | Available data set shows the expression changes in functionally related transcripts involved in ubiquitine proteosome, ribosome, sulfur and methionine metabolism, sulfur-related reactions, and DNA polymerase. |
---|
Proceeding: | 1. Uploading the data | After downloading the .xls file from the PNAS server we go to the main page of AYM and upload the .xls file in the "Upload Data or Model" dialogue. A file type does not need to be set and we click the "Submit" button. | |
|
---|
2. Selecting names | We are now looking at the contents in our .xls file. In the top row we can select the different sheets in the file. Since our dataset contains only one sheet, we can ignore this row. Now we are allowed to select a column or row containing the names of the genes we are interested in. Therefore, we click on the blue "3" in the second row, which means that we are selecting the third column. | |
|
---|
3. Annotating the data | To annotate our data, we move our mouse over the "Annotate Automatically" button in the top right column on the main screen. We could select databases which should be used for the annotation process or restrict the assigned annotations to those relevant to the model finding process by checking the "Only search-relevant annotations" box and finally hit the "go" button, or we simply use all web resources and click the "Annotate Automatically" button. (For all examples below we have annotated all elements. Annotating only those elements which are similar to elements in the BioModels database speeds up the search process, but it can also lead to slightly different results: the scores will be higher and more significant and depending on the size of the investigated models also the order in the retrieval might change.) | |
|
---|
4. Investigating annotation | The automatically assigned annotations can be reviewed and refined in the "Annotations" frame. By clicking e.g. on "UBP14" we can look at the annotations in the right frame. We can delete and add annotations here and search for new annotations via the name of a gene. Furthermore, we can remove entries in the left panel by clicking on the "stop" sign, e.g. for "Gene name". The green flags in the left panel show that the data is almost completely annotated. | |
|
---|
5. Find similar models | Now we can search for models containing the same or similar annotations by clicking the "Find Similar Models" button in the top right panel. | |
|
---|
6. Similar models | The results of the retrieval of similar models show models of amino acid metabolism and ubiquitination. This is in good agreement with the articles findings. | |
|
---|
7. Local alignment | We can have a close look at the results by clicking on the magnifier icon, e.g. for BioModel 66. It clearly shows that model and data are overlapping in the aspartate/homoserine kinase and the aspartate-semialdehyde/homoserine dehydrogenase. | |
|
---|
8. Visual alignment | After increasing the "Splitting nodes with degree bigger than" number, checking the "Show graph" box, and submitting the form we get a nice force-directed graph of the model overlaid with the data. Red and blue nodes distinguish the data and the model. In the model circles represent species while squares represent reactions while green, red, and blue edges connect substrates, products, and modifiers to reactions. | |
|
---|
|
Data set Moxley et al (click to expand) | Publication title | Linking high-resolution metabolic flux phenotypes and transcriptional regulation in yeast modulated by the global regulator Gcn4p |
---|
Authors | Moxley, Jewett, Antoniewicz, Villas-Boas, Alper, Wheelerd, Tong, Hinnebusch, Ideker, Nielsen, and Stephanopoulos |
---|
Supplementary material | Dataset 1 and 2 |
---|
Results | Quantification of more than 5700 mRNAs in yeast under stress with or without the global regulator Gcn4p trying to predict metabolic responses in amino acid pathways. |
---|
Proceeding: | 1. CAUTION | The first data set of this example does not work with a slow connection to our server. If you get timeouts trying to work with this data set, either use the "Only search-relevant annotations" option while annotating the data automatically or please download our web server and run it locally on your machine. | |
|
---|
2. Finding models in accordance with the mRNAs | Similar to the above mentioned steps, we load the .xls file and select the first column of the "mRNA_chemostat". In the annotate panel we delete the first 9 entries which are no Yeast ORFs and press the "Annotate automatically" button. Finally, we hit the "Find similar models" button. | |
|
---|
3. Analysing hits | The local alignments of the hits to the data show general annotations, such as "cAMP phosphodiesterase", "cAMP dependent protein kinase", "phosphoproteine phosphatase", "protein tyrosine phosphatase", and "ATP:protein phosphotransferase". | |
|
---|
4. Finding models in accordance with the metabolic data | We load data set 2 and select the first column from the sheet "HPLC Metabolites". Then we delete the first seven entries and annotate the remaining entries using KEGG by hovering the mouse over the "Annotate Automatically", checking the "KEGG Compound" box, and clicking "go". | |
|
---|
5. Delete incorrect annotations | The annotation of "Met" has been assigned slightly incorrectly. Therefore, we delete its annotation by selecting "Met" in the left panel and clicking on "del" beside the one present annotation. | |
|
---|
6. Add annotations by hand | We select the entries "His", "Met", and "Pro" in the left panel. For each entry we search for the annotations "l-histidine", "l-methionine", and "l-proline", respectively, and add the "KEGG Compound" annotation by clicking on the "add" button right to the annotation. | |
|
---|
7. Finding similar models | After clicking on the "Find similar models" button, we get a model set of 12 models describing metabolism of amino acids. | |
|
---|
8. Investigating hits | The first model contains 7 amino acids, as it can be seen after clicking the local alignment button of BioModel 212, while the last model an the list (number 90) contains only "l-cysteine". | |
|
---|
|
Data set Kresnowati et al (click to expand) | Publication title | When transcriptome meets metabolome: fast cellular responses of yeast to sudden relief of glucose limitation |
---|
Authors | Kresnowati, van Winden, Almering, ten Pierick, Ras, Knijnenburg, Daran-Lapujade, Pronk, Heijnen, and Daran |
---|
Supplementary material | Supplementary data 2 and 7 |
---|
Results | Changes in metabolite levels and transcriptional reprogramming in yeast after relief from glucose limitation. |
---|
Proceeding: | 1. CAUTION | The first data set of this example does not work with a slow connection to our server. If you get timeouts trying to work with this data set, either use the "Only search-relevant annotations" option while annotating the data automatically or please download our web server and run it locally on your machine. | |
|
---|
2. Finding models similar to the transcriptional data | We upload the first data set (number 2), select column 2 of the only sheet, annotate everything automatically, and hit the find similar models button. The results are very similar to those of the first Moxley data set. The problem here are the very general annotations which are matched before those included in central carbon metabolism. | |
|
---|
3. Finding models similar to the metabolic data | We upload the second data set (number 7), select the second row of the "Intracellular metabolites" sheet, and annotate all entries by hand with ChEBI IDs. Alternatively, we could download the file http://semanticsbml.org/aym/static/kresnowati7.csv and upload this file instead. Then we search for similar models and find models describing central carbon metabolism: TCA cycle and glycolysis. | |
|
---|
4. Local alignments | The overlap between the data and the models are as expected metabolites of TCA cycle or glycolysis. | |
|
---|
|
Data set of Brauer et al (click to expand) | Publication title | Conservation of the metabolomic response to starvation across two divergent microbes |
---|
Authors | Brauer, Yuan, Lu, Kimball, Botstein, and Rabinowitz |
---|
Supplementary material | Complete metabolite data |
---|
Results | Metabolites in E. coli and yeast changing after carbon and nitrogen starvation. |
---|
Proceeding: | 1. Finding similar models | We upload the data set, select the first column of the "CDWnormed all" sheet, and annotate the entries automatically using ChEBI annotations. Some more annotations have to be added by hand. Alternatively, we could upload the annotation file at http://semanticsbml.org/aym/static/brauer.csv . After clicking on the "Find similar models" button we are presented the following results. Most models are concerned with glycolysis and amino acid metabolism, but some of them just overlap in phosphonucleotides or NADH. | |
|
---|
2. Local alignment | E.g. BioModel 172 overlapping in the above mentioned substances and phosphoenolpyruvate, trehalose, and succinate. | |
|
---|
|
Data set of Feist et al (click to expand) | Publication title | A genome-scale metabolic reconstruction for Escherichia coli K-12 MG1655 that accounts for 1260 ORFs and thermodynamic information |
---|
Authors | Feist, Henry, Reed, Krummenacker, Joyce, Karp, Broadbelt, Hatzimanikatis, and Palsson |
---|
Supplementary material | Supplementary Information 1 |
---|
Results | Reconstructed model of E. coli K-12 MG1655 metabolism. |
---|
Proceeding: | 1. Finding similar models | We upload the data set, select column 2 (official reaction names) from the "reactions_GPRs" sheet, annotate the entries automatically, and search for similar models. The result list contains as expected models of carbon and amino acid metabolism. | |
|
---|
2. Local alignment | Especially interesting is the relatively high similarity score of BioModel 212. This score might be due to the fact that we have annotated the data mainly with EC numbers and GeneOntology annotations and that the annotations in model 212 (as in model 15) are to a relatively large extend EC numbers. | |
|
---|
|
|