Merge branch 'CW-2891' into 'dev'

control for sample sheet

Closes CW-2891

See merge request epi2melabs/workflows/wf-transcriptomes!139

CHANGELOG.md

@@ -4,6 +4,9 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [unreleased]
+- Sample sheet must include a `control` type to indicate which samples are the reference for the differential expression pipeline.
+
 ## [v0.4.1]
 ### Changed
 - Updated docker container with Pychopper to support LSK114.

README.md

@@ -190,16 +190,16 @@ Differential Expression requires at least 2 replicates of each sample to compare
 The sample sheet should be a comma separated values file (.csv) and include at least three columns named `barcode`, `alias` and `condition`.
 - Each `barcode` should refer to a directory of the same name in the input FASTQ directory (in the example below `barcode01` to `barcode06` reflect the `test_data` directory).
 - The `alias` column allows you to rename each barcode to an alias that will be used in the report and other output files.
-- The condition column will need to contain one of two keys to indicate the two samples being compared. for example: treated/untreated, sample/control etc.
+- The condition column will need to contain one of two keys to indicate the two samples being compared. Control must be one of the keys, used to indicate which samples will be used as the reference in the differential expression analysis.
 
 In the default `sample_sheet.csv` available in the test_data directory we have used the following.
 
 eg. sample_sheet.csv
 ```
 barcode,alias,condition
-barcode01,sample01,untreated
-barcode02,sample02,untreated
-barcode03,sample03,untreated
+barcode01,sample01,control
+barcode02,sample02,control
+barcode03,sample03,control
 barcode04,sample04,treated
 barcode05,sample05,treated
 barcode06,sample06,treated

bin/de_analysis.R

@@ -19,7 +19,11 @@ cts <- as.matrix(read.csv("merged/all_counts.tsv", sep="\t", row.names="Referenc
 coldata <- read.csv("de_analysis/coldata.tsv", row.names="alias", sep=",", stringsAsFactors=TRUE)
 
 coldata$sample_id <- rownames(coldata)
-coldata$condition <- factor(coldata$condition, levels=rev(levels(coldata$condition)))
+# check if control condition exists, sets as reference 
+if(!"control" %in% coldata$condition)
+  stop("sample_sheet.csv does not contain 'control' 
+       condition - unable to set reference.")
+coldata$condition <- relevel(coldata$condition, ref = "control")
 
 cat("Loading annotation database.\n")
 

bin/workflow_glue/check_sample_sheet_condition.py

@@ -1,5 +1,6 @@
 #!/usr/bin/env python
 """Check if a sample sheet is valid."""
+from collections import Counter
 import csv
 import sys
 
@@ -11,30 +12,27 @@ def main(args):
     logger = get_named_logger("checkSheetCondition")
     with open(args.sample_sheet, "r") as f:
         csv_reader = csv.DictReader(f)
-        unique_controls = []
-        controls_dic = {}
+        conditions_count = Counter()
         for row in csv_reader:
-            if 'condition' in list(row.keys()):
-                unique_controls.append(row['condition'])
-                if row['condition'] not in controls_dic:
-                    controls_dic[row['condition']] = 1
-                else:
-                    controls_dic[row['condition']] += 1
+            if "condition" in row:
+                conditions_count[row['condition']] += 1
             else:
                 sys.exit(
                     "Sample sheet has no condition column "
                     "which is required for the "
                     "differential expression subworkflow.")
-        if len(list(set(controls_dic.keys()))) != 2:
+        if len(conditions_count.keys()) != 2:
             sys.exit(
                 "There must be only two unique conditions "
                 "in the condition column of the sample sheet.")
-        for val in list(controls_dic.values()):
-            if val < 2:
-                sys.exit(
-                    "There must be at least 2 repeats for each "
-                    "condition indicated in the sample sheet.")
-
+        if "control" not in conditions_count:
+            sys.exit(
+                "One of the condition types must be control, "
+                "to indicate which samples to use as the reference.")
+        if any(v < 2 for v in conditions_count.values()):
+            sys.exit(
+                "There must be at least 2 repeats for each "
+                "condition indicated in the sample sheet.")
     logger.info(f"Checked sample sheet for condition column {args.sample_sheet}.")
 
 

bin/workflow_glue/tests/test_check_sample_sheet_condition.py

@@ -10,6 +10,7 @@ ERROR_MESSAGES = [
     ("sample_sheet_1.csv", "There must be only two unique conditions in the condition column of the sample sheet."),  # noqa: E501
     ("sample_sheet_2.csv", "Sample sheet has no condition column which is required for the differential expression subworkflow."),  # noqa: E501
     ("sample_sheet_3.csv", "There must be at least 2 repeats for each condition indicated in the sample sheet."),  # noqa: E501
+    ("sample_sheet_4.csv", "One of the condition types must be control, to indicate which samples to use as the reference."),  # noqa: E501
 ]
 
 

docs/quickstart.md

@@ -115,16 +115,16 @@ Differential Expression requires at least 2 replicates of each sample to compare
 The sample sheet should be a comma separated values file (.csv) and include at least three columns named `barcode`, `alias` and `condition`.
 - Each `barcode` should refer to a directory of the same name in the input FASTQ directory (in the example below `barcode01` to `barcode06` reflect the `test_data` directory).
 - The `alias` column allows you to rename each barcode to an alias that will be used in the report and other output files.
-- The condition column will need to contain one of two keys to indicate the two samples being compared. for example: treated/untreated, sample/control etc.
+- The condition column will need to contain one of two keys to indicate the two samples being compared. Control must be one of the keys, used to indicate which samples will be used as the reference in the differential expression analysis.
 
 In the default `sample_sheet.csv` available in the test_data directory we have used the following.
 
 eg. sample_sheet.csv
 ```
 barcode,alias,condition
-barcode01,sample01,untreated
-barcode02,sample02,untreated
-barcode03,sample03,untreated
+barcode01,sample01,control
+barcode02,sample02,control
+barcode03,sample03,control
 barcode04,sample04,treated
 barcode05,sample05,treated
 barcode06,sample06,treated

nextflow_schema.json

@@ -97,10 +97,10 @@
             "properties": {
                 "sample_sheet": {
                     "type": "string",
-                    "title": "Sample sheet",
+                    "title": "Sample and condition sheet",
                     "format": "file-path",
-                    "description": "A CSV file used to map barcodes to sample aliases. The sample sheet can be provided when the input data is a directory containing sub-directories with FASTQ files. If you are running the differential expression workflow, there should be an additional column `condition` with any two distinct labels eg. `treated`,`untreated`. There should be at least 3 repeats for each condition.",
-                    "help_text": "The sample sheet is a CSV file with, minimally, columns named `barcode` and `alias`. Extra columns are allowed. A `type` column is required for certain workflows and should have the following values; `test_sample`, `positive_control`, `negative_control`, `no_template_control`."
+                    "description": "A CSV file used to map barcodes to sample aliases. The sample sheet can be provided when the input data is a directory containing sub-directories with FASTQ files. If you are running the differential expression workflow, there must be an additional column `condition` with two labels, one of which must be `control` (e.g. `control` and `treated`). Control will indicate which samples will be used as the reference. There should be at least 3 repeats for each condition.",
+                    "help_text": "The sample sheet is a CSV file with, minimally, columns named `barcode` and `alias`. Extra columns are allowed."
                 },
                 "sample": {
                     "type": "string",

test_data/sample_sheet.csv

@@ -1,7 +1,7 @@
 barcode,sample_id,alias,condition
-barcode01,sample01,sample01,untreated
-barcode02,sample02,sample02,untreated
-barcode03,sample03,sample03,untreated
+barcode01,sample01,sample01,control
+barcode02,sample02,sample02,control
+barcode03,sample03,sample03,control
 barcode04,sample04,sample04,treated
 barcode05,sample05,sample05,treated
 barcode06,sample06,sample06,treated

test_data/workflow_glue/check_sample_sheet_condition/sample_sheet_1.csv

@@ -1,7 +1,7 @@
 barcode,sample_id,alias,condition
-barcode01,sample01,sample01,untreated
-barcode02,sample02,sample02,untreated
-barcode03,sample03,sample03,untreated
+barcode01,sample01,sample01,control
+barcode02,sample02,sample02,control
+barcode03,sample03,sample03,control
 barcode04,sample04,sample04,treated
 barcode05,sample05,sample05,treated
 barcode06,sample06,sample06,other

test_data/workflow_glue/check_sample_sheet_condition/sample_sheet_3.csv

@@ -1,3 +1,3 @@
 barcode,sample_id,alias,condition
-barcode01,sample01,sample01,untreated
+barcode01,sample01,sample01,control
 barcode04,sample04,sample04,treated

test_data/workflow_glue/check_sample_sheet_condition/sample_sheet_4.csv

@@ -0,0 +1,7 @@
+barcode,sample_id,alias,condition
+barcode01,sample01,sample01,untreated
+barcode02,sample02,sample02,untreated
+barcode03,sample03,sample03,untreated
+barcode04,sample04,sample04,treated
+barcode05,sample05,sample05,treated
+barcode06,sample06,sample06,treated