Repair opencga-storage-mongodb variants database due to opencb/opencga#504

repair_mongodb_opencga_504.js

// Generic migration utility method
function migrateCollection(collection, query, projection, migrateFunc) {
    var bulk = db.getCollection(collection).initializeOrderedBulkOp();
    var count = 0;
    var bulkSize = 500;
    db.getCollection(collection).find(query,projection).forEach(function(doc) {

        migrateFunc(bulk, doc);

        if ( bulk.nUpdateOps > bulkSize ) {
            count += bulk.nUpdateOps;
            print("Execute bulk! " + count);
            bulk.execute();
            bulk = db.getCollection(collection).initializeOrderedBulkOp();
        }

    });

    if ( bulk.nUpdateOps > 0 ) {
        count += bulk.nUpdateOps;
        print("Execute bulk! " + count);
        bulk.execute();
        bulk = db.getCollection(collection).initializeOrderedBulkOp();
    }
    if (count == 0) {
        print("Nothing to do!");
    }

}

// Remove extra secondary alternate and fix type on variants with reference.length == 50
migrateCollection("variants", {reference:{$regex:"^.{50}$"}, type:"SV"}, {stats:0, annotation:0}, function (bulk, doc) {
    var ref = doc.reference;
    var id = doc._id;
    print("Updating variant: _id : '" + id + "'");

    for (var i in doc.studies) {
        var study = doc.studies[i];
        bulk.find({_id: id, "studies.sid": study.sid}).update({$set: {type: "INDEL"}, $pull: {"studies.$.alts": {ref: ref}}});
    }

});

// Change type of SV secondary alternates with reference.length = 50
migrateCollection("variants", {"studies.alts":{$elemMatch:{type:"SV", ref:{$regex:"^.{50}$"}}}}, {stats:0, annotation:0}, function (bulk, doc) {

    var id = doc._id;
    print("Updating variant: _id : '" + id + "'");

    for (var i in doc.studies) {
        var study = doc.studies[i];
        if (study.hasOwnProperty("alts")) {
            var alts = study.alts;
            var modified = false;
            for (var i2 in alts) {
                var alt = alts[i2];
                if (alt.ref.length == 50 && alt.type == "SV") {
                    alt.type = "INDEL";
                    modified = true;
                }
            }
            if (modified) {
                bulk.find({_id: id, "studies.sid": study.sid}).update({$set: {"studies.$.alts": alts}});
            }
        }
    }
});