Merge pull request #606 from hechth/hechth/issue603

Updated mfassignr to version 1.1.2
RECETOX · Nov 1, 2024 · 7d98b78 · 7d98b78
2 parents 5da5b5e + f515aa6
commit 7d98b78
Show file tree

Hide file tree

Showing 8 changed files with 103 additions and 121 deletions.
diff --git a/tools/mfassignr/macros.xml b/tools/mfassignr/macros.xml
@@ -1,5 +1,5 @@
 <macros>
-    <token name="@TOOL_VERSION@">1.1.1</token>
+    <token name="@TOOL_VERSION@">1.1.2</token>
     <xml name="requirements">
         <requirements>
             <requirement type="package" version="@TOOL_VERSION@">r-mfassignr</requirement>
@@ -208,26 +208,6 @@
        </section>
     </xml>
 
-    <xml name="findrecalseries_param">
-        <param name="input_file" type="data" format="tabular" label="Input data"
-               help= "Recalibration series, output from RecalList"/>
-        <param name="global_min" type="float" label="Global min" value="0" min="0"
-               help= "A lower bound of the instrument m/z range."/>
-        <param name="global_max" type="float" label="Global max" value="1000" min="0"
-               help= "A higher bound of the instrument m/z range."/>
-        <param name="number_of_combinations" type="integer" label="Number of combinations"
-               help= "Combinations of how many series should be computed." value="5"/>
-        <param name="abundance_score_threshold" type="float" label="Abundance score threshold" value="100"
-               help= "A threshold for filtering abundance score parameter. The series with higher values are better."/>
-        <param name="peak_distance_threshold" type="float" label="Peak distance threshold" value="2"
-               help= "A threshold for the peak distance parameter. The closer this value is to 1, the better."/>
-        <param name="coverage_threshold" type="integer" label="How many % of the m/z range should be covered."
-               help= "How many % of the m/z range should be covered." value="90"/>
-        <param name="fill_series" type="boolean" truevalue="TRUE" falsevalue="FALSE" label="Whether to return top 10 unique series (TRUE) or series only from the best combination."
-               help= "If TRUE, top 10 unique series will be returned, otherwise only the series from the best
-               combination will be returned." value="false"/>
-    </xml>
-
     <xml name="recal_param">
         <param name="input_file" type="data" format="tabular" label="Input data (Output from MFAssign)"
                help= "Input data frame, the output from MFAssign or MFAssignCHO"/>

diff --git a/tools/mfassignr/mfassignr_findRecalSeries.xml b/tools/mfassignr/mfassignr_findRecalSeries.xml
@@ -12,53 +12,44 @@
     </edam_operations>
     <expand macro="creator" />
     <expand macro="refs" />
-
     <expand macro="requirements" />
+
     <command detect_errors="exit_code"><![CDATA[
         Rscript -e 'packageVersion("MFAssignR")' &&
         Rscript '${mfassignr_findrecalseries}'
     ]]>
     </command>
+
     <configfiles>
         <configfile name="mfassignr_findrecalseries"><![CDATA[
             library(dplyr)
             df <- read.table("$input_file", header=TRUE, sep="\t")
-            result <- MFAssignR::FindRecalSeries(
-                df = df,
-                global_min = $global_min,
-                global_max = $global_max,
-                number_of_combinations = $number_of_combinations,
-                abundance_score_threshold = $abundance_score_threshold,
-                peak_distance_threshold = $peak_distance_threshold,
-                coverage_threshold = $coverage_threshold,
-                fill_series = $fill_series
-            ) |> dplyr::rename(Series = series)
-            
+            result <- MFAssignR::FindRecalSeriesSimple(df) |> dplyr::slice(-1)
             write.table(result, file="$final_series", sep="\t", row.names=FALSE)
         ]]></configfile>
     </configfiles>
+
     <inputs>
-        <expand macro="findrecalseries_param"/>
+        <param name="input_file" type="data" format="tabular" label="Input data"
+        help="Recalibration series, output from RecalList"/>
     </inputs>
+
     <outputs>
         <data name="final_series" format="tabular" label="Final recalibration series"/>
     </outputs>
+
     <tests>
         <test>
             <param name="input_file" value="recallist/recal_series.tabular"/>
-            <param name="global_min" value="100"/>
-            <param name="global_max" value="500"/>
-            <param name="abundance_score_threshold" value="50"/>
-            <param name="number_of_combinations" value="3"/>
-            <param name="coverage_threshold" value="50"/>
-            <param name="fill_series" value="TRUE"/>
-            <output name="final_series" file="findrecalseries/selected_series.tabular"/>
+            <output name="final_series" file="findrecalseries/selected_series_simple.tabular"/>
         </test>
     </tests>
+
     <help>
         @FINDRECALSERIES_HELP@
 
         @GENERAL_HELP@
     </help>
+
     <expand macro="citations"/>
 </tool>
diff --git a/tools/mfassignr/mfassignr_recal.xml b/tools/mfassignr/mfassignr_recal.xml
@@ -71,13 +71,13 @@
             <param name="isopeaks" value="isofiltr/iso_out.tabular"/>
             <output name="Mono">
                 <assert_contents>
-                    <has_n_lines n="9203"/>
+                    <has_n_lines n="9156"/>
                     <has_n_columns n="3"/>
                 </assert_contents>
             </output>
             <output name="Iso">
                 <assert_contents>
-                    <has_n_lines n="2474"/>
+                    <has_n_lines n="2521"/>
                     <has_n_columns n="4"/>
                 </assert_contents>
             </output>

diff --git a/tools/mfassignr/mfassignr_snplot.xml b/tools/mfassignr/mfassignr_snplot.xml
@@ -20,8 +20,8 @@
             df = read.delim("$input_file", sep="\t"),
             cut = cut,
             mass = $mass,
-            window.x = $window_x,
-            window.y = $window_y
+            mass_window = $window_x,
+            abundance_window = $window_y
         )
 
         ggplot2::ggsave(filename = "SNplot.png", SNplot)

diff --git a/tools/mfassignr/test-data/findrecalseries/selected_series_simple.tabular b/tools/mfassignr/test-data/findrecalseries/selected_series_simple.tabular
@@ -0,0 +1,11 @@
+"Series"	"Number.Observed"	"Series.Index"	"Mass.Range"	"Tall.Peak"	"Abundance.Score"	"Peak.Score"	"Peak.Distance"	"Series.Score"	"Num_Weight"	"Abund_weight"	"TallP_Weight"	"PeakS_Weight"	"PeakD_Weight"	"Series_Weight"	"Total_score"
+"O_H_7"	12	39	"131.049-299.236"	131.049175570564	172.333212266955	0.840925082711506	1	1.1	25	20	4.36830585235213	5.79537458644247	7	9.16666666666667	71.3303471054613
+"O7_H_4"	11	40	"233.066-415.268"	359.206083268496	-11.5251682369853	0.136136287521764	1	1.3	22.9166666666667	-1.33754464219376	11.9735361089499	9.31931856239118	7	8.125	57.9969766958139
+"O2_H_11"	5	78	"209.06-391.263"	223.074691790747	132.732152227999	0.540579659159889	1	2.8	10.4166666666667	15.4041290685615	7.43582305969157	7.29710170420056	7	6.08695652173913	53.6406770208594
+"O6_H_7"	9	50	"253.071-407.242"	351.1796217	11.6976374801684	0.213201464774084	2	1.3	18.75	1.35756042915837	11.70598739	8.93399267612958	4	8.125	52.872540495288
+"O10_H_3"	7	59	"381.175-465.268"	409.206584131861	2.6997318710302	0.0702879797813004	2	1	14.5833333333333	0.313315330865898	13.640219471062	9.6485601010935	4	10	52.1854282363548
+"O10_H_4"	9	47	"309.081-491.284"	435.222060757962	4.7511651696338	0.655346847143645	2	1.6	18.75	0.551392863526961	14.5074020252654	6.72326576428178	4	7.27272727272727	51.8047879258014
+"O9_H_6"	10	44	"303.07-499.289"	415.195395703443	-21.7793592335313	0.166622042493807	3	1.5	20.8333333333333	-2.52758698651699	13.8398465234481	9.16688978753096	1	7.5	49.8124826577954
+"O8_H_1"	5	69	"213.061-311.169"	311.16936123622	27.8251864824375	0.198145538844287	1	1.6	10.4166666666667	3.22923087388803	10.3723120412073	9.00927230577856	7	7.27272727272727	47.3002091602679
+"O9_H_9"	7	62	"339.071-479.226"	367.101864488525	-18.5226214411156	0.0271291741355916	2	1.6	14.5833333333333	-2.14962875669292	12.2367288162842	9.86435412932204	4	7.27272727272727	45.8075147949739
+"O10_H_6"	6	66	"319.065-487.254"	431.190612941439	-34.0254140710475	0.137560981917351	1	2.2	12.5	-3.94879357536027	14.3730204313813	9.31219509041325	7	6.47058823529412	45.7070101817284