1.5.0: Rank now based on score, evalue and profile name

conda/meta.yaml

@@ -1,4 +1,4 @@
-{% set version = "1.4.3" %}
+{% set version = "1.5.0" %}
 
 package:
   name: hmmrank

src/R-test/Makefile

@@ -4,7 +4,7 @@ CHECK = if [ ` $(DIFF) | wc -l ` -eq 0 ]; then echo "*** $@ OK ***"; else echo "
 
 all: hmmrank
 
-hmmrank: hmmrank.00 hmmrank.01 hmmrank.02 hmmrank.03 hmmrank.04 hmmrank.05 hmmrank.06 hmmrank.07 hmmrank.08 #hmmrank.09
+hmmrank: hmmrank.00 hmmrank.01 hmmrank.02 hmmrank.03 hmmrank.04 hmmrank.05 hmmrank.06 hmmrank.07 hmmrank.08 hmmrank.10
 
 hmmrank.00:
 	../R/hmmrank.r --outfile=$@.out $@.d/*.tblout
@@ -46,6 +46,12 @@ hmmrank.08:
 	@$(CHECK)
 
 # I was missing entries with the classical accept score table
+# NOT FINISHED -- DOESN'T WORK
 hmmrank.09:
 	../R/hmmrank.r --annottable=$@.annottable.tsv --minscore=30 --maxscore=100 --scorefile=$@.d/profile_accept_scores.tsv --outfile=$@.out $@.d/*.tblout
 	@$(CHECK)
+
+# Duplicate bitscores
+hmmrank.10:
+	../R/hmmrank.r --only_best_scoring --outfile=$@.out $@.d/*.tblout
+	@$(CHECK)

src/R-test/hmmrank.06.expect

@@ -66,7 +66,7 @@ envision.uc0.99_ENV3_S3_k141_286858_3_1	PF06180	3.2e-6	34.3	26.2	1	All	CbiK	4.99
 envision.uc0.99_ENV3_S6_k141_296524_2_1	PF06180	2.5e-5	31.4	26.2	1	All	CbiK	4.99.1.3	T
 envision.uc0.99_ENV3_S6_k141_343739_1_1	PF02572	3.8e-57	200.9	25.7	1	All	BtuR	2.5.1.17	T
 testing_presence_in_all_TIGRFAMS	TIGR01466 & TIGR01467	1e-10	200	200	1	All	CobIJ	2.1.1.130 & 2.1.1.131	T
-testing_presence_in_all_TIGRFAMS	TIGR01466	1e-10	100	100	3	All	CobI	2.1.1.130	T
+testing_presence_in_all_TIGRFAMS	TIGR01466	1e-10	100	100	2	All	CobI	2.1.1.130	T
 testing_presence_in_all_TIGRFAMS	TIGR01467	1e-10	100	100	3	All	CobJ	2.1.1.131	T
-testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	100	3	All	CysG	2.1.1.107	F
-testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	100	3	All	CobA	2.1.1.107	F
+testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	100	4	All	CysG	2.1.1.107	F
+testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	100	4	All	CobA	2.1.1.107	F

src/R-test/hmmrank.07.expect

@@ -45,7 +45,7 @@ envision.uc0.99_ENV3_S3_k141_286858_3_1	PF06180	3.2e-6	34.3	26.2	1	All	CbiK	4.99
 envision.uc0.99_ENV3_S6_k141_296524_2_1	PF06180	2.5e-5	31.4	26.2	1	All	CbiK	4.99.1.3	T
 testing_presence_in_all_TIGRFAMS	TIGR01466 & TIGR01467	1e-10	200	200	1	Bacteria	CobIJ	2.1.1.130 & 2.1.1.131	T
 testing_presence_in_all_TIGRFAMS	TIGR01466 & TIGR01467	1e-10	200	200	1	Archaea	CobIJ	2.1.1.130 & 2.1.1.131	T
-testing_presence_in_all_TIGRFAMS	TIGR01466	1e-10	100	100	3	All	CobI	2.1.1.130	T
+testing_presence_in_all_TIGRFAMS	TIGR01466	1e-10	100	100	2	All	CobI	2.1.1.130	T
 testing_presence_in_all_TIGRFAMS	TIGR01467	1e-10	100	100	3	All	CobJ	2.1.1.131	T
-testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	100	3	All	CysG	2.1.1.107	F
-testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	100	3	All	CobA	2.1.1.107	F
+testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	100	4	All	CysG	2.1.1.107	F
+testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	100	4	All	CobA	2.1.1.107	F

src/R-test/hmmrank.08.expect

@@ -45,7 +45,7 @@ envision.uc0.99_ENV3_S3_k141_286858_3_1	PF06180	3.2e-6	34.3	26.2	1	All	CbiK	4.99
 envision.uc0.99_ENV3_S6_k141_296524_2_1	PF06180	2.5e-5	31.4	26.2	1	All	CbiK	4.99.1.3	T
 testing_presence_in_all_TIGRFAMS	TIGR01466 & TIGR01467	1e-10	200	409.6	1	Bacteria	CobIJ	2.1.1.130 & 2.1.1.131	T
 testing_presence_in_all_TIGRFAMS	TIGR01466 & TIGR01467	1e-10	200	409.6	1	Archaea	CobIJ	2.1.1.130 & 2.1.1.131	T
-testing_presence_in_all_TIGRFAMS	TIGR01466	1e-10	100	246.35	3	All	CobI	2.1.1.130	T
+testing_presence_in_all_TIGRFAMS	TIGR01466	1e-10	100	246.35	2	All	CobI	2.1.1.130	T
 testing_presence_in_all_TIGRFAMS	TIGR01467	1e-10	100	163.25	3	All	CobJ	2.1.1.131	T
-testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	185	3	All	CysG	2.1.1.107	F
-testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	185	3	All	CobA	2.1.1.107	F
+testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	185	4	All	CysG	2.1.1.107	F
+testing_presence_in_all_TIGRFAMS	TIGR01469	1e-10	100	185	4	All	CobA	2.1.1.107	F

src/R/hmmrank.r

@@ -14,7 +14,7 @@ suppressPackageStartupMessages(library(tidyr))
 suppressPackageStartupMessages(library(data.table))
 suppressPackageStartupMessages(library(stringr))
 
-SCRIPT_VERSION = '1.4.3'
+SCRIPT_VERSION = '1.5.0'
 
 # Get arguments
 # For interactive testing:
@@ -25,6 +25,9 @@ SCRIPT_VERSION = '1.4.3'
 #
 # TC06:
 # opt <- list('options' = list('verbose' = TRUE, qfromfname = TRUE, minscore = 30, maxscore=100, scorefile = 'hmmrank.06.profile_scores.tsv', annottable = 'hmmrank.06.annottable.tsv'), 'args' = Sys.glob('hmmrank.06.d/*.tblout'))
+#
+# TC10 (ties):
+# opt <- list('options' = list('verbose' = TRUE, qfromfname = FALSE, minscore = 0), 'args' = Sys.glob('hmmrank.10.d/*.tblout'))
 option_list = list(
   make_option(
     '--annottable', type = "character",
@@ -175,14 +178,18 @@ if ( length(opt$options$annottable) > 0 ) {
     filter(identical(sort(strsplit(profilecomb, '\\s*&\\s*')[[1]]), sort(strsplit(profiles, '\\s*&\\s*')[[1]]))) %>%
     ungroup() %>%
     group_by(accno) %>% 
-    mutate(rank = rank(desc(score))) %>% 
+    arrange(desc(score), evalue, profiles) %>%
+    mutate(rank = row_number()) %>%
     ungroup() %>%
     transmute(accno, profile = profilecomb, evalue, score, min_score, rank) %>%
     inner_join(annottable, by = 'profile')
 } else {
   # We didn't have an annotation table, just calculate ranks
   tblout <- lazy_dt(tblout) %>%
-    group_by(accno) %>% mutate(rank = rank(desc(score))) %>% ungroup() %>%
+    group_by(accno) %>% 
+    arrange(desc(score), evalue, profile) %>%
+    mutate(rank = row_number()) %>%
+    ungroup() %>%
     as_tibble()
 }