annotate usearch-wrapper.sh @ 7:b6ece07bec6a draft

planemo upload commit 45906ba522c7c319067e93d5cd5d7161223c3336
author bcclaywell
date Mon, 12 Oct 2015 15:59:59 -0400
parents d4690e65afcd
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
1 #!/bin/bash
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
2
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
3 source $(dirname $0)/util.sh
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
4 source $1
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
5
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
6 RDP_SEQS="/shared/silo_researcher/Matsen_F/MatsenGrp/micro_refset/rdp/10_31/tax_filter/filtered/rdp_10_31.filter.fasta"
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
7 RDP_SEQINFO="/shared/silo_researcher/Matsen_F/MatsenGrp/micro_refset/rdp/10_31/tax_filter/filtered/rdp_10_31.filter.seq_info.csv"
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
8
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
9 sqlite3 -csv -header ${CLASS_DB} <<EOF > usearch_meta.csv
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
10 SELECT pn.name, CAST(pn.mass AS INT) count, tax_id, tax_name, taxa.rank
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
11 FROM multiclass_concat
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
12 JOIN taxa USING (tax_id)
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
13 JOIN placement_names pn USING (placement_id, name)
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
14 WHERE want_rank = 'species';
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
15 EOF
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
16
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
17 romp -v usearch_clusters \
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
18 --usearch-quietly \
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
19 --query-group tax_id \
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
20 --query-duplication count \
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
21 --database-name seqname \
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
22 --database-group tax_id \
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
23 ${INPUT_SEQS} \
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
24 usearch_meta.csv \
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
25 ${RDP_SEQS} \
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
26 ${RDP_SEQINFO} \
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
27 ${USEARCH_HITS} \
d4690e65afcd Uploaded
bcclaywell
parents:
diff changeset
28 ${USEARCH_GROUPS}