add support for UniProt variants annotations

[fkaiserbio]

• required adaptions: implement XML parsing in UniProtContentHandler, add new feature class
• we should add the support for genetic variants mapping in the UniProt database
• extraction of dbSNP identifier from description attribute
• for this the parsing of an additional feature is necessary:

  <feature type="sequence variant" description="In AMYL-TTR; amyloid polyneuropathy; dbSNP:rs121918091." id="VAR_007575" evidence="25 57 70">
  <original>F</original>
  <variation>L</variation>
  <location>
  <position position="84"/>
  </location>
  </feature>

• along with this, parsing of literature evidence should be implemented, specified by the evidence attribute of the "sequence variation" feature:

  <reference key="1">
  <citation type="journal article" date="1984" name="Biochem. Biophys. Res. Commun." volume="124" first="558" last="564">
  <title>
  Cloning and sequence analysis of cDNA for human prealbumin.
  </title>
  <authorList>
  ...
  </reference>

  f.e. evidence="1 2" would refer to references 1 and 2. For now, the parsing of the doi should be sufficient.

• for a good example see P02766

[cleberecht]

implemented support for sequence variants in commit 4fd7d42 (branch feature/mutation-variation-parsing)

• added new feature class SequenceVariants
• the following features of sequence variations are extracted: id, description, evidence, original, variation, and location
• dbSNP is not always annotated in description, neither is description in general. If you explicitly require it I would recommend implementing an Identifier (e.g. see ECNumber), registering it in the IdentifierPatternRegistry and adding the corresponding attribute to the SequenceVariant class
• evidences are parsed, previous FeatureOrigins are now reimplemented as Evidence, containing the evidence type, name (AuthorYear if literature), and preferably DOI, if no DOI is available pubmed or nothing otherwise

You can get the variations using:

Protein transthyretin = UniProtParserService.parse("P02766");
List<SequenceVariant> sequenceVariants = transthyretin.getFeature(SequenceVariants.class).getFeatureContent();

Please report errors and bugs in this thread.

[fkaiserbio]

• seems to work partly, test SingleFileChannelMembraneTransportTest fails
• it seems that there are some index access issues when parsing the AminoAcidFamily from the sequence by its one-letter code

[cleberecht]

should be fixed in bf99b60

[fkaiserbio]

merged in development https://github.com/singa-bio/singa/commit/093fca42cb31ffac09b01ca911a737b5ae1dca8c