mzueva / blast-wrapper Goto Github PK

Currently we don't have appropriate validation for the incoming parameters for task with type BLAST_TOOL.
Let's implement it in the following way:

Current schema of the input parameters for BLAT_TOOL task is:

{
  "blastTool": "string",
  "algorithm": "string",
  "dbName": "string",
  "taxIds": [
    long
  ],
  "excludedTaxIds": [
    long
  ],
  "query": "string",
  "maxTargetSequence": 0,
  "expectedThreshold": 0,
  "options": "string"
}

Regarding to this schema we need to check that:

• blastTool is one of blastn, blastp, blastx, tblastx, tblastn
• algoritm should be provided only for blastn blastp with possible values
  • blastn: megablast, dc-megablast, blastn
  • blastp: blastp, 'blastp-fast, blastp-short
  • blastx: blastx blastx-fast
  • tblastn: tblastn tblastn-fast
• Verify as it is now
• taxIds and excludedTaxIds - should not be present with each other
• query - verify as it is now
• maxTargetSequence - lets introduce new application prop value that would be a hard top limit, maxTargetSequence should be > 0 and < that this limit
• expectedThreshold check that this value is > 0
• options - here we need to introduce approach to verify and remove bad options, lets for now implement this in the following way
  • check that parameter -num_threads is not provided, if provided remove it with its value
  • check that parameter -outfmt is not provided, if provided remove it with its value
  • check that parameter -remote is not provided, if provided remove it with its value
  • check that parameter -negative_taxids, -negative_taxidslist is not provided, if provided remove it with its value
  • check that parameter -taxids -taxidslist is not provided, if provided remove it with its value
  • check that parameter -evalue is not provided, if provided remove it with its value
  • check that parameter -out is not provided, if provided remove it with its value
  • check that parameter -db is not provided, if provided remove it with its value
  • check that parameter -task is not provided, if provided remove it with its value
  • check that parameter -query is not provided, if provided remove it with its value
  • check that parameter -num_alignments is not provided, if provided remove it with its value
  • check that parameter -num_descriptions is not provided, if provided remove it with its value

Would be great to have possibility to download a raw blast result as a file through a Rest API

Right now we already have approach for blast-wrapper to work as cluster: #17
However there is several artifacts:

• Any host in a cluster could start any job from a queue
• It isn't possible to discern task that are running on other host right now and task that hanged after host is failed

It would be grate to solve this problems and adds optional possibility to run specific tasks on specific hosts

We need to provide possibility to get results of the specific BLAST_TOOL task.

• Let's create a Rest API GET /blast/{id} method to get a result
  • possible parameter: size > 0 to limit number of returned result entries

There should be a way to cancel currently running task. Such API should be available through the Rest API.

Let's implement this in the following way:

• provide Rest API GET /task/{id}/cancel method to cancel a task with id={id}
• If such method was called find such task and if it still in progress - cancel a Future
• Regarding to task type schedule a new task for clean up process (deleting results files etc)

Currently blast is executed with number of customize parameters such as

• blastTool
• db
• query

This list should be extended regarding to the schema of BlastStartSearchingRequest:

{
  "blastTool": "string",
  "algorithm": "string",
  "dbName": "string",
  "taxIds": [
    long
  ],
  "excludedTaxIds": [
    long
  ],
  "query": "string",
  "maxTargetSequence": 0,
  "expectedThreshold": 0,
  "options": "string"
}