TargetScan 7.0 information


Changes from version 6.2 to version 7.0 (August 2015)


Changes from version 6.1 to version 6.2 (April 2012)


Changes from version 6.0 to version 6.1 (March 2012)


Changes from version 5.2 to version 6.0 (November 2011)

TargetScanHuman and Mouse were improved and updated:

  1. An updated set of RefSeq genes and their annotations was used to define a set of human or mouse 3' UTRs.
  2. All 3' UTRs (rather than just the longest one) were used for miRNA target prediction.
  3. Orthologous UTRs for the same set of 23 species as version 5 were obtained using updated 46-way multiz (whole-genome) alignments from UCSC Genome Bioinformatics.
  4. An updated set of miRNAs and miRNA families, based on miRBase Release 17, was used for target prediction. The classification of broadly conserved and conserved remains the same as TargetScan 5.2, but many new poorly conserved miRNA families have been added.
  5. Context+ scores (Garcia et al., 2011) were used in place of context scores. These expanded scores include the same four contributions as before and also
  6. TargetScanWorm and TargetScanFly remain the same.

Changes from version 5.1 to version 5.2 (June 2011)

  1. TargetScanWorm was improved and updated:
  2. For TargetScanFly ORFs (released in August 2010), miRNA sites were predicted in ORFs, in addition to 3' UTRs.
  3. TargetScanHuman and TargetScanMouse remain the same.

Changes since the release of version 5.1

  1. For TargetScanFly (all versions), miRNA family miR-2a-1/6/11/13/306 has been renamed to miR-2a-1/6/11/13/308.

Changes since version 5.0

  1. TargetScanHuman and TargetScanMouse remain the same except for the addition of a link to a gene's 3' UTR in the UCSC Genome Browser of the reference species (human and mouse, respectively).
  2. For TargetScanWorm and TargetScanFly, predicted consequential pairing to 3' end of miRNAs is only shown if the raw 3' pairing score (Grimson et al., 2007) is at least 3.0. This removes some presumably inconsequential pairing displayed in previous versions.

Changes from version 4.2 to version 5.0 (December 2008)

TargetScanFly and TargetScanWorm remain the same.

The mammalian implementation of TargetScan was overhauled using methods and insights described in Friedman et al., 2009 (Genome Research, 2009) to create TargetScanHuman and TargetScanMouse.

  1. This release introduces the TargetScanMouse database, which uses mouse-centric 3' UTR annotations.
  2. An updated set of miRNAs and miRNA families, based on miRBase Release 10.1, was used for target prediction.
  3. The classification of miRNA families has been updated to partition them more accurately into broadly conserved, conserved, and poorly conserved groups.
  4. An updated set of RefSeq genes and their annotations was used to define a set of human or mouse 3' UTRs. Orthologous UTRs (based on whole-genome alignments) were obtained for 22 other species from UCSC Genome Bioinformatics, thereby extending the analyses beyond the original five species.
  5. Conservation of each miRNA site was evaluated using phylogenetic branch lengths of all species containing the site. Substantially improved methods for estimating chance conservation, which also control for differential local conservation rates, were used (Friedman et al., 2009).
  6. For all highly conserved miRNAs, the probability of preferentially conserved targeting (PCT) for each site was estimated as described (Friedman et al., 2009).
  7. Grouping of aligned sites was made more flexible to detect sites in the reference species that are conserved as shorter sites in other species (Friedman et al., 2009).
  8. Predicted targeting involving 3'-compensatory sites (Friedman et al., 2009) has been included in TargetScanHuman.
  9. Tabular output showing conserved predicted targets of a miRNA family has been extended to permit ranking by either total context score or aggregate PCT.
  10. Target prediction performed without considering site conservation is made more accessible, with the tabular output showing both conserved and nonconserved targets ranked by total context score available as an option.
  11. Tabular output showing predicted targets includes those predicted by TargetScan 4 as "2007" in "Previous TargetScan publication(s)".
  12. Yellow highlighting in the graphical view of a UTR has been modified to appear around sites in which PCT > 0.75.
  13. Predicted consequential pairing to 3' end of miRNAs is only shown if the raw 3' pairing score (Grimson et al., 2007) is at least 3.0. This removes some presumably inconsequential pairing displayed in previous versions.

Changes from version 4.1 to version 4.2 (April 2008)

  1. For TargetScanFly, target mRNAs from chromosome X genes, which were inadvertently omitted from previous releases, are included.
  2. For TargetScanFly, new predictions were generated using a revised set of 3' UTRs that were based on Flybase 5.4 annotations. Where a transcript had no annotated 3' UTR, its 3' UTR was defined as the region downstream of the last coding exon, 1500 nt long or 20 nt beyond the nearest AAUAAA, or to the nearest sequence of any adjacent protein-coding or RNA gene, whichever was shorter.
  3. For TargetScan vertebrate, fly, and worm datasets, since overlapping sites for each miRNA family cannot be occupied at the same time, a set of only non-overlapping sites is shown in the summary table of miRNA targets. For vertebrates, the most favorable total context score that considers only non-overlapping sites is shown. All overlapping sites are still shown on the graphical view.
  4. For TargetScanFly and TargetScanWorm, predicted targets of a miRNA family are sorted by the number and type of sites, with genes listed in decreasing total site weight. Site types are weighted using the site-type score extrapolated from experiments on vertebrate cells (Grimson et al., 2007), in which 8mer = 0.31, 7mer-m8 = 0.161 and 7mer-A1 = 0.099. Poorly conserved sites are given the same weight as conserved sites except in the case of ties. Overlapping sites to the same miRNA were handled as described in change (3).

Changes from version 4.0 to version 4.1 (December 2007)

  1. The major change is an expansion of TargetScanFly to include miRNAs newly identified by Ruby et al., 2007 (Genome Res., 448:83).
  2. Context scores for mammalian sites that start 15 nt from a stop codon are now considered. (Version 4.0 only considered scores for mammalian sites greater than 15 nt from the stop codon.)
  3. The end coordinates of 8mer mammalian sites on UTRs have been corrected by 1 nt. The previous coordinates had altered the AU-content score, and these scores have now been corrected.
  4. Scores for 7mer-A1 mammalian sites have been adjusted by -0.032. Previous scores were too high because of a typo that has been corrected in Table S6 of Grimson et al., 2007.
  5. TargetScanFly and TargetScanWorm now identify predicted pairing between an miRNA and its target gene as is Grimson et al., 2007 (like mammalian TargetScan).
  6. MicroRNA family miR-200b/429 has been renamed to miR-200bc/429.
  7. MicroRNA family miR-124.2/506 has been renamed to miR-124/506 and has been expanded to include hsa-mir-124.
  8. Rare cases of incorrect mapping of TargetScan sites to the human genome have been fixed.
  9. Predicted targets are now displayed at the UCSC genome browser (Human, March 2006).

Changes from version 3.1 to version 4.0 (July 2007):

  1. Additional features have been considered to better predict target specificity, as detailed in Grimson et al., 2007. Sites within 15 nt of a stop codon are flagged because these are typically not effective. The context of each of the remaining sites has been evaluated and scored considering the following four features:
    With all four features, a more negative score is associated with a more favorable site. The context score is the sum of the above scores, and the context score percentile is the percentile rank of each site compared to all sites for this miRNA family. Thus a high context score percentile (between 50 and 100) shows that a specific site is more favorable than most other sites of this miRNA. In a gene with multiple sites for one miRNA family, a total context score is calculated as the sum of context scores for the most favorable (most negative) miRNA in this family (as shown on the miRNA family pages [example]). If the context score for any of these sites is positive (unfavorable), its contribution to the total context score is '0'. The "representative miRNA" is the miRNA in this family with the most favorable total context score.

  2. Mammalian miRNA families have been updated to include recent additions to miRBase. The 3' UTR dataset has been updated to include chicken and is now based on current RefSeq annotation and genome coordinates.

  3. Degree of conservation, for both miRNA families and predicted targets, has been expanded to include three levels:
    UTR consensus sequence (shown below each UTR alignment on gene-centric pages [example]) is uppercase at "highly conserved" positions and lowercase at "conserved" positions.

    Changes from version 3.0 to version 3.1