Data file(s) *

One or more data files. Each file represents one gene expression dataset (expression profiles of a set of genes over a set of conditions).

If multiple data files are provided, they will be clustered simultaneously. These multiple datasets can be from different experements, labs, sets of conditions, and even different species. However, if you wish to analyse each dataset independently, submit each dataset in a separate job.

Each dataset file is a tab, space, or comma delimited file with columns representing samples and rows representing genes. The first column in the file must have gene names/IDs and the first row must be a header with column names (i.e. titles of the samples). Check this link for further details.

((close))

You have selected multiple datasets to be clustered simultaneously. If these datasets use different gene IDs (e.g. different microarray probe IDs or orthologous genes from different species), please provide a gene ID map file.

Gene ID map file for multiple datasets (optional)

The gene ID map file defines which gene IDs from one dataset correspond to which gene IDs from the other datasets. If your datasets come from different species, check this link for full description of the gene ID map file. If your datasets were generated by different technologies (e.g. mixing RNA-seq with microarrays or using different microarray platforms), check this link for full description of the gene ID map file.

((close))

Minimum datasets for gene inclusion

The minimum number of datasets in which a gene should be represented for it to be included in the analysis. This should be a number between 1 and the number of submitted datasets.

((close))

Normalisation

Select one of these data normalisation options:

Option	Description
Automatic normalisation (default)	Clust automatically detects the required normalisation for each dataset and applies it. This is recommended if unsure which normalisation to choose.
Normalisation codes	You specify which normalisation techniques to apply by providing space-split normalisation codes. For example: 101 3 4. If you submitted multiple datasets, each will be normalised using the same techniques. Full list of codes is available here.
Normalisation file	Provide a file that describes what normalisation techniques should be applied to the dataset(s). This is can be used to specify different normalisation techniques for different datasets. Full description of the format of the file is available here.
No normalisation (not recommended)	This should not be chosen unless the data has already been normalised in a way that suits clustering. Hint: data normalised by differential expression analysis tools do not tend to be ready for clustering as clustering requires different types of normalisation.

((close))

Codes

Replicates structure (optional)

A file that lists every group of samples that represent replicates of the same condition. The format of this file is explained here.

((close))

Cluster tightness (t)

This is a real positive number that controls clusters' tightness. The default is 1.0. For tighter clusters, use bigger values (e.g. 2, 5, or 10), and for less tight clusters use smaller values (e.g. 0.5 or 0.1).

((close))

Minimum cluster size (cs)

Filter-out flat genes before clustering (Recommended)

If checked, genes with flat expression profiles (no change of expression over all conditions) are filtered out. This includes filtering out genes with zero expression values in all conditions.

((close))

Filter-out low expression genes before clustering (Recommended)

Filter-out genes with low expression values. When clicked, more parameters appear to be edited.

((close))

Filter out genes which do not exceed the expression percentile of:

At least in this number of conditions:

At least in this number of datasets:
(Only relevant if multiple datasets were submitted)

Data is two sided (e.g. log-ratios)

If checked, filtering will consider large positive values as valuable as large negative values (e.g. for log-ratios)

((close))

Use expression percentiles

If checked, filtering threshold will be based on the percentile of all gene expression values in the datasets instead on actual expression values. In this case, provide a value between 0 and 100 in the relevant text-box above.

((close))

Email address (optional)

If provided, a notification will be sent to this email once the job is done.

((close))