Launch prediction server Query precalculated results
Algorithms Prediction server Database query Output description Benchmark results

Description of the input items for querying pre-calculated results

  1. Choose the virus and type of sequence. Currently either nucleotide or protein sequence can be selected for HIV-1 or HCV.
  2. The sequences are periodically fetched from NCBI GenBank/GenPept and the databases are updated accordingly.
  3. This table lists the statistics of genotyping results in comparison with those from Los Alamos (LANL) database. 'Agree' means the number of cases MuLDAS and LANL agree to each other, while 'Disagree' indicates otherwise. For viruses showing recombinant patterns such as HIV-1, subtypings are performed hierarchically, first using 'major' groups and followed by 'nested' analysis. If two analyses do not agree, MuLDAS employs a complex decision rules as described in (http://www.muldas.org/side.cgi?FILENAME=benchmark.results.html). Based on these rules, sequences are hierarchically categorized from I through III. The leftovers are labeled as 'Others', for which MuLDAS withheld predicting genotype/subtype. It includes ¡®Problematic¡¯ sequences, which cannot be genotyped with high confidence (P < 0.9) nor located close to any genotype (Outlierness > 10). One can list the sequences in each category by clicking the relevant links.
  4. The database entries can be queried by a list of NCBI GIs or accession codes. The items in the list can be separated by white spaces (blank, tab, or newline).
  5. Optionally one can test the system using the exemplary entries given here.
  6. The query can be restricted to the cases of a certain predicted subtype / genotype.
  7. The query can be restricted to the cases of a certain range of posterior probabilities.
    The query can be restricted to the cases of a certain range of outlierness values.
    The query can be restricted to the cases of a certain range of miss-classification rates.
  8. The query can be restricted to the cases of a certain gene.
  • In the pre-calculation, the Jukes-Cantor correction is used to calculate the genetic distances and all MDS dimensionalities were 10. So if you want to view the result by the different options, please use the prediction server.

Copyrighted by Bio-Data Mining Lab, Department of Bioinformatics and Life Sciences, Soongsil University, Seoul, Korea