About:
DOUT score:

The D-sum score is a quantitative estimate rating the quality of a domain outlier (DOUT) prediction for a protein family F with multiple sub-significant hits of a domain D. The D-score interprets the sum-scores, Sr , for all occurrences, r, of a domain D within a family F normalized by λ, and penalizes for the size of the search space, mn, where n is the database length. The query length m, and the constants λ and κ are calculated for a concatenation product of all residues within family F. A reward proportional to the sum of the average domain coverage, Cr, and the ratio of domain instances to the number of all proteins, N, within the family is applied.
D-sum score cutoffs are set to 24 (estimated 5% false positive rate) and 13 (estimated 10% false positive rate) based on an analysis of subsignificant domain hits in ASTRAL families.

Domain ranking view:
The domain ranking view lists domains according to their supportive evidence in the following succession:

• » Domains with a significant (E<0.01) hit within the set.
  Please note that occassionaly rps-blast report false positive similarities with a E-value slightly below 0.01. If only one hit of a domain is found in the set and it has a significant E-value > 1e-05 a warning is issued (!!!: 1 hit only, low sim).

• » Domains with several subsignificant hits in the sequence set, which are considered probable distant domain homolog based on an estimated 5% false positive rate
• » Domains with several subsignificant hits in the sequence set, which are considered potential distant domain homolog based on an estimated 10% false positive rate
• » And other domains, which appear multiple times in the protein set

• » domain common name used as a link to a detailed summary table for all hits of this type in the set (column 1); the D-score is reported for the "probable" and "potential" classes
• » links to the NCBI conserved domain database (CDD), to the original domain databases (SMART, and Pfam) and Pfam-Clans (column 2)
• » the number of domain hits of this type within the set (sub- subsignificant hits, sig- significant hits), and the number of proteins (prot) in which these hits appear (column 3)
• » short information about the domain (column 4)

Details section:
In the details section the user can access:

• » The input sequences in a simple text file.
• » The masked and non-redundant input sequence set that was used for the domain analysis.
• » The rps-blast result of the masked and non-redundant input in simple text format.
• » A detailed tabular listing of the obtained domain hits.

Graphical section:
In the graphical domain view proteins are listed according to their input order. The following color-code is used in domain visualization:

• » Red: significant domains (E < 0.01 )

• » Orange: supported by a significant hit in the protein set
• » Blue: probable distant domain homolog based on an estimated 5% false positive rate
• » Cyan: potential similarity based on an estimated 10% false positive rate

Detailed tabular view:
In the detailed tabular view domains are listed according to the associated evidence as in the short domain ranking table, but here additional information on each scored domain hit is avaible in a html/text format. Sorting of the different tables according to E-value, score, start, end and coverage are enabled by clicking on the respective keywords.

Single Sequence input:
If a single sequences is provided an initial psi-blast search will be performed to collect a homologous sequence set. This initial analysis step will take several minutes. All consequent steps parallel the analysis of a homologous sequence set as described above.

Simple text file upload:
You can upload your sequences in a simple text format. To generate a simple text file you can use any text-processing software, and save file as text.

Rps-Blast flag: SEG Filter
You can determine whether rps-blast will be run with a SEG low complexity filter switched on (T) or off (F). If filtering is turned off, false positive hits can increase in compositionally biased regions.

Input flag: Coil Filter
You can determine whether a coiled-coil filter should be applied to your input (T) or not (F). If filtering is turned off, false positive hits can increase in compositionally biased regions.

Input flag: Transmembrane Filter
You can determine whether a HMMTOP-based filtering should be applied to your input (T) or not (F). If filtering is turned off, false positive hits can increase in compositionally biased regions.

Protein set flag: maximum-sequence identity
CD-HIT is used to obtain a non-redundant protein set with a user determined identity cut-off. The set is made non-redundant, in order to reduce noise due to highly similar sequences.

DOUT-analysis flag: Expect
Subsignificant domain hits (>0.01) are only taken into consideration as potential domain outliers in case the E-value is below this user defined threshold. False positive results are rare with the default setting of 0.01. Higher E-value thresholds give more false positives, while lower E-values increase reliability.

DOUT-analysis flag: Coverage
Subsignificant domain hits (>0.01) are only taken into consideration as potential domain outliers in case the coverage of the domain is above this user-defined threshold.

Single seq flag: Blastpgp rounds
If a single seqeuence is given as an input PSI-BLAST is used to obtain a sequence set. You can change the maximum number of PSI-BLAST passes to use in multipass version.

Single seq flag: Inclusion threshold
If a single seqeuence is given as an input PSI-BLAST is used to obtain a sequence set. You can change the E-value threshold for inclusion in this initial search.

Single seq flag: Database choise
If a single seqeuence is given as an input PSI-BLAST is used to obtain a sequence set. You can run the PSI-BLAST against two versions of the NCBI non-redundant database (mar06), which have both processed using cd-hit. nr80d is a 80% non-redundant derivate of NCBI nr supplemented by Pfam 19 seed files, and the CDD Smart and Pfam domain fasta files. nr90d is a 90% non-redundant derivate of NCBI nr supplemented by Pfam 19 seed files, and the CDD Smart and Pfam domain fasta files.