Casella di testo: Tutorial –1- BATS input file format
Index - 1  - 2 - 3 - 4 - 5 - 6 - 7 – <previous – next>

The data file should be contained in a .xls or .txt tab delimited file prepared as follows:

 

The first row should contain in the first column a text string and in the remaining columns the numeric values of the time variable, for all the samples in the same unit (either seconds, hours, days, etc.), with the samples ordered in ascending order.

The first column (from the second row on) should contain gene identifiers: identifiers are unique strings (characters or combinations of characters and numbers).  

The remaining cells will contain the data in the form of log-scale signal to reference ratios. Data have to be already normalized.

Missing values can be denoted either by empty cell or with NaN 

 

A user friendly software for Bayesian Analysis of Time Series Microarray Experiments.

Bayesian Analysis for Time Series

Microarray Experiments

Gene Name or Gene Id

Casella di testo: Times (in increasing order) included replicates

Data (normalized in log2 scale) and with empty cell or NaN for missing values

Esplosione 1: DATA HAVE TO BE ALREADY NORMALIZED
Fumetto 2: Warnings: In current version genes with completely blank lines (i.e., all record for the a given gene are NaN) should be avoided, to ensure the files are correctly read. Please always check your input data file and use the filter data in order to remove genes with too many missing. We suggest to include a gene in the analysis only if at least 60-50% of the observations are not missing. You can use the utility filter data for achieve such conditions
Casella di testo:  Warnings: 
BATS implements a truly functional approach, hence by construction it is designed for those time-course experiments where at least 5-6 time points are available, although in order to fully exploit the advantage of any functional method a larger number of time points and of arrays is recommendable. From our studies we found that the availability of about 8-10 time points already provides a satisfactory analysis in most of the case. Similar requirements are however typical of any functional based methods.  For helping the user in the analysis a warning message will be displayed if the data-set does not meet some minimum constraint. Nevertheless, BATS can be still applied for analyzing shorter time course studies, but in this case no particular gain is obtained with respect to simpler regression approaches.
In principle all the  probes available on the arrays can be analyzed. The results of the analysis up to 50.000 probes and 25 arrays is usually returned in 10-20 minutes depending on the number of total probes, the  number of arrays available, the distributions of missing data and the configuration of the computer on which the analysis is performed. However from practical point of view, probes containing too many missing values should be removed from the analysis since they may not carry trustable information, similarly control probes or not expressed probes can be removed if their information is not considered significant or of biological interest.