# HG changeset patch # User greg # Date 1502296846 14400 # Node ID 3b27bfc37b83784ec1e429bfe789020c1404a0e0 Uploaded diff -r 000000000000 -r 3b27bfc37b83 ideas.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/ideas.xml Wed Aug 09 12:40:46 2017 -0400 @@ -0,0 +1,331 @@ + + accounts for position dependent epigenetic events and detects local cell type relationships + + bedtools + ucsc-bedgraphtobigwig + ucsc-bedsort + ucsc-bigwigaverageoverbed + ideas + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +**What it does** + +Employs the IDEAS (Integrative and Discriminative Epigenome Annotation System) method for jointly and quantitatively characterizing +multivariate epigenetic landscapes in many cell types, tissues or conditions. The method accounts for position dependent epigenetic +events and detects local cell type relationships, which not only help to improve the accuracy of annotating functional classes of DNA +sequences, but also reveal cell type constitutive and specific loci. The method utilizes Bayesian non-parametric techniques to automatically +identify the best model size fitting to the data so users do not have to specify the number of states. On the other hand, users can +still specify the number of states if desired. + +----- + +**Required options** + + * **Cell type, Epigenetic factor and Input** - specify any number of inputs with currently supported formats, either bam or bigwig. The cell name + factor name must be unique for each input. For example, if you have replicate data you may want to specify the cell name as "cell_rep1", "cell_rep2", etc and the factor name as "factor_rep1", "factor_rep2", etc. + + * **Cell type name** - cell type name + * **Epigenetic factor name** - epigenetic factor name + * **BAM or BigWig file** - BAM or BigWig file + + * **Set genomic windows on which to process the data** - if "No" is selected, IDEAS will run whole genome segmentation. If "Yes" is selected, IDEAS will segment genomes in the unit of the windows defined by the bed file. This file can be in BED3, BED4 or BED5 format, but only the first three columns (chr posst posed) will be used. + + * **Window size in base pairs** - Window size in base pairs (if "No" is selected) + * **Restrict processing to specified chromosomes** - If "Yes" is selected, processing will be restricted to specified chromosomes + + * **Chromosomes** - processing will be restricted to specified chromosomes (if "Yes" is selected) + + * **Chromosome** - specified chromosome + + * **Bed file specifying the genomic windows** - bed file specifying the genomic windows (if "Yes" is selected) + +**Other options** + +* **Output chromosomes in seperate files** - select "Yes" to produce seperate files for each chromosome, allowing you to run IDEAS on different chromosomes separately. +* **Select Bed file containing blacklist regions for exclusion** - select a Bed file that contains regions you'd like excluded from your datasets. +* **Standardize all datasets** - select "Yes" to standardize all datasets (e.g., reads / total_reads * 20 million) so that the signals from different cell types become comparable - your datasets can be read counts, logp-values or fold change. + +* **Discourage state transition across chromosomes** - select "Yes" to produce similar states in adjacent windows, making the annotation smoother, but at risk of reducing precision. +* **Run IDEAS only within defined windows in the input data** - select "Yes" to Run IDEAS only in windows between zero-based start and end indexes of windows in the input data. +* **Perform Log2-transformation of the input data** - select "Yes" to perform Log2-transformation of the input data by log2(x+1) (recommended for read count data to reduce skewness). You can optionally enter a number less than 1 to direct IDEAS to produce log2(x+num) transformation. For example, if your input data is mean read count per window, then 1 may be too large, but using 0.1 may be more reasonable. +* **Set the maximum number of states to be inferred** - select "Yes" to restrict the maximum number of states to be generated by IDEAS; the final number of inferred states may be smaller than the number you specified +* **Set the initial number of states** - select "Yes" if the number of states you expect to generate is greater than 20. While IDEAS may infer 30 states or more by starting from just 20 states, it may not do so if it is trapped in a local mode. We recommend setting the initial number of states slightly larger than the number of states you expect. +* **Maximum number of position classes to be inferred** - Set this value only if: + + * you do not want position classes (e.g., for testing purposes), in this case set the value t0 1 + * IDEAS runs slow because there are too many position classes, generally less than 100 position classes will run fine + +* **Maximum number of cell type clusters allowed** - Set this value only for testing. If you set the value to 1, then all cell types will be clustered in one group. +* **Prior concentration** - specify the prior concentration parameter; default is A=sqrt(number of cell types). A smaller concentration parameter (e.g., 1 or less) will emphasize more on position specificity and a larger concentration parameter (e.g., 10 * number of cell types) will emphasize more on global homogeneity. +* **Set the the number of burnin and maximization steps** - specify the number of burnin and maximization steps; default it is 50 50. Increasing these two numbers will increase computing and only slightly increase accuracy. Decreasing these two numbers will reduce computing but may also reduce accuracy. We recommend to run IDEAS with at least 20 burnins and 20 maximizations. IDEAS will not stop even if it reaches a maximum mode. +* **Set the minimum standard deviation for the emission Gaussian distribution** - specify the minimum standard deviation for the emission Gaussian distribution. + + * **Minimum standard deviation for the emission Gaussian distribution** - you should change the default minerr value of 0.5 if the standard deviation of your data is much smaller or much larger than 1. The first line of the output produced by IDEAS is **ysd=xxx**, which is the total standard deviation of your data. If that value is less than 0.5, you may set the minimum standard deviation to an even smaller number (e.g., xxx/2). If the standard deviation of your data is much greater than 1, (e.g., 20), you may set the minimum standard deviation to a larger value, (e.g., 5). Modifying the minimum standard deviation in the former case is more necessary than in the latter case because otherwise you may end up finding no interesting segmentations. We do not recommend setting the minimum standard deviation to be 0 or smaller, as doing so may capture some artificial and uninteresting states due to tightly clustered data, such as 0 in read counts. + +* **Set the maximum standard deviation for the emission Gaussian distribution** - specify the maximim standard deviation for the emission Gaussian distribution. + + * **Maximim standard deviation for the emission Gaussian distribution** - if you want to find fine-grained states you may use this option (if not used, IDEAS uses infinity), but it is rearely used unless you need more states to be inferred. + + + + 10.1093/nar/gkw278 + +