PhyloSkims/annotate
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Changes to be committed:

Browse Source 
modified:   .gitignore
	new file:   data/cds/sp_chlorodb/parameters.sh
	deleted:    data/ir/LSC_RefDB.fasta
	deleted:    data/ir/SSC_RefDB.fasta
	modified:   detectors/cds/bin/go_cds.sh
	modified:   detectors/normalize/lib/lookforIR.lib.sh
	modified:   detectors/normalize/lib/selectIR.py
	modified:   organnot/Dockerfile
	new file:   organnot/README.md
	new file:   organnot/dorgannot
	deleted:    ports/.DS_Store
	deleted:    src/ncbiblast/binaries/.gitignore
	deleted:    src/prokov/lxpack/tests/S.fasta
	deleted:    src/prokov/lxpack/tests/St.fasta
	deleted:    src/prokov/lxpack/tests/Stt.fasta
	deleted:    src/prokov/lxpack/tests/aS.fasta
	deleted:    src/prokov/lxpack/tests/aSt.fasta
	deleted:    src/prokov/lxpack/tests/aStt.fasta
	deleted:    src/prokov/lxpack/tests/aaS.fasta
	deleted:    src/prokov/lxpack/tests/aaSt.fasta
	deleted:    src/prokov/lxpack/tests/aaStt.fasta
	new file:   src/repseek/repseek-2014.09.tgz
This commit is contained in:
Eric Coissac
2025-03-05 21:56:39 +01:00
parent 5d09615f27
commit 4b71fe8c4c
22 changed files with 94 additions and 75646 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
detectors/cds/bin/go_cds.sh
View File

@ -26,7 +26,7 @@ needarg 1
Fasta=$1; shift
needfile $Fasta
needfile "$Fasta"
# Genome names is set from the base
# name of the genome file without its extension

6
detectors/normalize/lib/lookforIR.lib.sh
View File

@ -27,9 +27,9 @@ function lookForIR {
-outfmt 6 \
-max_target_seqs 10000 | \
awk -v id_match=80 -v lmin=100 \
'($4 > lmin) && (($3+0)>id_match) {
SAME=(($7 < $8) && ($9 < $10)) || (($7 > $8) && ($9 > $10));
if ($7 < $8)
'($4+0 > lmin) && (($3+0)>id_match) {
SAME=(($7+0 < $8+0) && ($9+0 < $10+0)) || (($7+0 > $8+0) && ($9+0 > $10+0));
if ($7+0 < $8+0)
{print substr($2,1,3),$7,$8,SAME}
else
{print substr($2,1,3),$8,$7,SAME}

30
detectors/normalize/lib/selectIR.py
View File

@ -50,13 +50,15 @@ for line in data:
# <Zafacs> 07/13/2023
# Hack for avoiding crash when LSC and SSC have no blast similarity
# Need to be reworked
if (len(chloro['SSC']) > 0) :
maxSSC = float(max(abs(n) for n in chloro['SSC']))
chloro['SSC']=[n / maxSSC for n in chloro['SSC']]
if (len(chloro['SSC']) > 0) :
sSSC = sum(abs(x) for x in chloro['SSC'])
# maxSSC = float(max(abs(n) for n in chloro['SSC']))
chloro['SSC']=[n / sSSC for n in chloro['SSC']]
if (len(chloro['LSC']) > 0) :
maxLSC = float(max(abs(n) for n in chloro['LSC']))
chloro['LSC']=[n / maxLSC for n in chloro['LSC']]
sLSC = sum(abs(x) for x in chloro['LSC'])
# maxLSC = float(max(abs(n) for n in chloro['LSC']))
chloro['LSC']=[n / sLSC for n in chloro['LSC']]
scoreMax=0
len1Max=0
@ -105,17 +107,19 @@ for line in repeats:
c = float(c_lsc + c_ssc)
o = float(o_lsc + o_ssc)
if c > 0:
c_lsc /= c
c_ssc /= c
# if c > 0:
# c_lsc /= c
# c_ssc /= c
if o > 0:
o_lsc /= o
o_ssc /= o
# if o > 0:
# o_lsc /= o
# o_ssc /= o
score = ((c_lsc - c_ssc) ** 2 + (o_lsc - o_ssc) ** 2) / 2.0
#score = ((c_lsc - c_ssc) ** 2 + (o_lsc - o_ssc) ** 2) / 2.0
score = abs(abs(c_lsc) + abs(o_ssc) - abs(o_lsc) - abs(c_ssc))
# pvalue=
# print >>sys.stderr,"c.lsc = %f c.ssc = %f o.lsc = %f o.ssc = %f score = %6.4f (len=%d)" % (c_lsc,c_ssc,o_lsc,o_ssc,score,len1)
# print("c.lsc = %f c.ssc = %f o.lsc = %f o.ssc = %f score = %6.4f (len=%d)" % (c_lsc,c_ssc,o_lsc,o_ssc,score,len1),
# file=sys.stderr)
if (score >= scoreMax) and ((len1 > len1Max) or (len2 > len2Max)):
scoreMax = score