On 2018 Jan 24, Jean-Michel Claverie commented:

A brand new version of the above statistical test is now available: ACD 2.0

When the initial version of this test was published, transcriptome data was painfully obtained from "Expressed sequenced tags (EST)" library sequencing, resulting in low counts for each detected transcript. Thanks to the evolution in sequencing technologies ("NGS"), transcriptomes are now investigated using several hundred millions of reads, with every transcripts been detected up to several thousands of times.

A new (free) web service is now available that can handle all levels of counts, from a handful to millions, without approximation and without loosing the mathematical simplicity and universality of the original Audic-Claverie Distribution (ACD) test.

ACD 2.0 now proposes three tools:

1) the simple "one item /2 counts" --> p-value of the null hypothesis (i.e. no change in proportions)

2) "an array of items/ 2 or more counts" --> generate a ranked list of the most discriminant items

3) "an array of items /2 or more counts" --> generate a pairwise distance matrix of the whole samples

A full documentation describes the mathematical details and the computational algorithms used in ACD 2.0. It also explains how the above tools can be used in many more contexts than just transcriptome analysis. These include the comparison of metagenomic/barcoding or ChIP-Seq experiments, or non-biological applications simply involving arrays of items and their cognate counts. A formal publication will follow soon. Keep posted in PubMed!

Without further delay you can start using ACD 2.0 (beta) from the following the link:

http://www.igs.cnrs-mrs.fr/acdtool

On 2018 Jan 24, Jean-Michel Claverie commented:

A brand new version of the above statistical test is now available: ACD 2.0

When the initial version of this test was published, transcriptome data was painfully obtained from "Expressed sequenced tags (EST)" library sequencing, resulting in low counts for each detected transcript. Thanks to the evolution in sequencing technologies ("NGS"), transcriptomes are now investigated using several hundred millions of reads, with every transcripts been detected up to several thousands of times.

A new (free) web service is now available that can handle all levels of counts, from a handful to millions, without approximation and without loosing the mathematical simplicity and universality of the original Audic-Claverie Distribution (ACD) test.

ACD 2.0 now proposes three tools:

1) the simple "one item /2 counts" --> p-value of the null hypothesis (i.e. no change in proportions)

2) "an array of items/ 2 or more counts" --> generate a ranked list of the most discriminant items

3) "an array of items /2 or more counts" --> generate a pairwise distance matrix of the whole samples

A full documentation describes the mathematical details and the computational algorithms used in ACD 2.0. It also explains how the above tools can be used in many more contexts than just transcriptome analysis. These include the comparison of metagenomic/barcoding or ChIP-Seq experiments, or non-biological applications simply involving arrays of items and their cognate counts. A formal publication will follow soon. Keep posted in PubMed!

Without further delay you can start using ACD 2.0 (beta) from the following the link:

http://www.igs.cnrs-mrs.fr/acdtool