Nextflow tutorial

.gitignore

@@ -1,3 +1,7 @@
 *.fastq.gz
 .nextflow*
 **/work/*
+input/human_genome
+nextflow/kneaddata
+nextflow/tutorial-inprogress.nf
+input/humann_databases/full_chocophlan.v201901_v31.tar.gz

components.typ

@@ -19,7 +19,9 @@
 
   #v(2em)
 
-  Kevin Bonham, PhD
+  Kevin Bonham \
+  Sagun Maharjan \
+  Emily Green
 
   2025-07-11
 ]
@@ -30,14 +32,15 @@
 1. Tasks (processes, rules, etc): single step in a workflow (typically one command)
   - inputs
   - outputs
-  - command
+  - command#pause
 2. Workflow: Order of steps (instructions to build computational graph)
+  - Also a way to find / specify inputs#pause
 3. Parameters: variables that may change from run to run
   - may be global (eg output directory)
   - or task-specific (eg memory to allocate)
 
 
-== Workflows builds computational graph
+== Workflows build computational graph
 
 #let bent-edge(from, to, ..args) = {
   let midpoint = (from, 50%, to)
@@ -50,10 +53,12 @@
   edge(..vertices, "-|>", ..args)
 }
 
-#set text(16pt)
+#text(16pt)[
 #diagram(
-  node-stroke: luma(80%),
+  node-stroke: blue,
   edge-corner-radius: none,
+  edge-stroke: purple.darken(50%),
+  label-sep:0.2em,
   spacing: (10pt, 47pt),
 
   // Nodes
@@ -67,7 +72,7 @@
   node((0,2), [Task 3 output 1], name: <d>),
   node((1,2), [Task 3 output 2], name: <e>),
 
-  node((2.5,2), [Task3 output], name: <f>),
+  node((2.5,2), [Task 4 output], name: <f>),
 
   node((3.5,3), [Report], name: <h>),
 
@@ -76,17 +81,225 @@
 
   node((8,2), [outputs...], name: <u>),
 
+  edge((-0.3,3),(0.1,3), "-|>"),
+  node((0.4,3), [Task], stroke:none),
+  node((0,3.5), [], width:1em,height:1em),
+  node((0.4,3.5), [File ], stroke: none),
+
   // Edges
   bent-edge(<a>, <b>, [_Task 1_]),
   bent-edge(<a>, <c>, [_Task 2_]),
   bent-edge(<b>, <d>, [_Task 3_]),
   bent-edge(<b>, <e>),
   edge(<c>, <f>, "-|>", [_Task 4_]),
-  edge(<f>, <h>, "-|>"),
   edge(<e>, <h>, "-|>"),
-  edge(<d>, <h>, "-|>"),
   edge(<x>, <y>, "--|>", [Tasks...]),
   edge(<w>, <u>, "--|>", [Tasks...]),
   edge(<y>, <h>, "-|>"),
   edge(<u>, <h>, "-|>"),
 )
+]
+
+== In anadama, tasks are added to workflows
+
+#slide(composer: (40%,60%))[
+#set text(16pt)
+```python
+workflow.add_task(
+  "cat [depends[0]] > [targets[0]]",
+  depends = input_files,
+  targets = output_files
+)
+```
+
+][
+#set text(16pt)
+#diagram(
+  node-stroke: blue,
+  edge-corner-radius: none,
+  edge-stroke: purple.darken(50%),
+  label-sep:0.2em,
+  spacing: (10pt, 47pt),
+
+  // Nodes
+  node((1.5,0), [*input_file*], stroke:3pt, name: <a>),
+  node((0.5,1), [*output_file*], stroke:3pt, name: <b>),
+  node((2.5,1), [Task 2 output], name: <c>),
+  node((0,2), [Task 3 output 1], name: <d>),
+  node((1,2), [Task 3 output 2], name: <e>),
+
+  node((2.5,2), [Task 4 output], name: <f>),
+  bent-edge(<a>, <b>, [*_task1_*]),
+  bent-edge(<a>, <c>, [_Task 2_]),
+  bent-edge(<b>, <d>, [_Task 3_]),
+  bent-edge(<b>, <e>),
+  edge(<c>, <f>, "-|>", [_Task 4_]),
+
+  node(stroke:teal+2pt, enclose:(<a>,<b>))
+)
+]
+== In nextflow, tasks are `process`es
+
+#slide(composer: (40%,60%))[
+#set text(16pt)
+
+```groovy
+process task1 {
+  input:
+  path input_file
+
+  output:
+  path output_file
+
+  shell:
+  """
+  cat $input_file > $output_file
+  """
+}
+```
+
+][
+#set text(16pt)
+#diagram(
+  node-stroke: blue,
+  edge-corner-radius: none,
+  edge-stroke: purple.darken(50%),
+  label-sep:0.2em,
+  spacing: (10pt, 47pt),
+
+  // Nodes
+  node((1.5,0), [*input_file*], stroke:3pt, name: <a>),
+  node((0.5,1), [*output_file*], stroke:3pt, name: <b>),
+  node((2.5,1), [Task 2 output], name: <c>),
+  node((0,2), [Task 3 output 1], name: <d>),
+  node((1,2), [Task 3 output 2], name: <e>),
+
+  node((2.5,2), [Task 4 output], name: <f>),
+  bent-edge(<a>, <b>, [*_task1_*]),
+  bent-edge(<a>, <c>, [_Task 2_]),
+  bent-edge(<b>, <d>, [_Task 3_]),
+  bent-edge(<b>, <e>),
+  edge(<c>, <f>, "-|>", [_Task 4_]),
+
+  node(stroke:teal+2pt, enclose:(<a>,<b>))
+)
+]
+
+#slide(composer: (40%,60%))[
+
+#set text(16pt)
+```groovy
+process task3 {
+  input:
+  path in_from_1
+
+  output:
+  path out1
+  path out2
+
+  shell:
+  """
+  cat $input_file > $out1
+  echo "I'm done!" > $out2
+  """
+}
+```
+
+][
+#set text(16pt)
+#diagram(
+  node-stroke: blue,
+  edge-corner-radius: none,
+  edge-stroke: purple.darken(50%),
+  label-sep:0.2em,
+  spacing: (10pt, 47pt),
+
+  // Nodes
+  node((1.5,0), [Input file 1], name: <a>),
+  node((0.5,1), [*in_from_1*], name: <b>, stroke:3pt),
+  node((3.5,1), [Task 2 output], name: <c>),
+  node((0,2), [*out1*], name: <d>, stroke:3pt),
+  node((1.5,2), [*out2*], name: <e>, stroke:3pt),
+
+  node((3.5,2), [Task 4 output], name: <f>),
+  bent-edge(<a>, <b>, [_Task 1_]),
+  bent-edge(<a>, <c>, [_Task 2_]),
+  bent-edge(<b>, <d>, [*_task3_*], label-pos:0.6),
+  bent-edge(<b>, <e>),
+  edge(<c>, <f>, "-|>", [_Task 4_]),
+  node(stroke:teal+2pt, enclose:(<b>,<d>,<e>))
+)
+]
+
+== Processes are called like functions
+
+#slide(composer: (43%,50%))[
+#set text(15pt)
+```groovy
+workflow {
+  input_ch = Channel
+    .fromPath("inputs/*.txt")
+
+  t1_out = task1(input_ch)
+  t2_out = task2(input_ch)
+
+  t3_out = task3(t1_out)
+  t4_out = task4(t2_out)
+
+  report = report_task(t3_out
+    .collect().map {
+      t3-> t3[1]
+  })
+
+}
+```
+
+][
+#set text(13pt)
+#diagram(
+  node-stroke: blue,
+  edge-corner-radius: none,
+  edge-stroke: purple.darken(50%),
+  label-sep:0.2em,
+  spacing: (10pt, 47pt),
+
+  // Nodes
+  node((1.5,0), [input_ch[0]], name: <a>),
+  node((0.5,1), [t1_out], name: <b>),
+  node((2.5,1), [t2_out], name: <c>),
+  node((6, 0), [input_ch[n]], name:<w>),
+
+  node((5.5, 1), [...], stroke:none),
+
+  node((0,2), [t3_out[0]], name: <d>),
+  node((1,2), [t3_out[1]], name: <e>),
+
+  node((2.5,2), [t4_out], name: <f>),
+
+  node((2.5,3), [Report], name: <h>),
+
+  node((5,0), [input_ch[1]], name: <x>),
+  node((5,2), [outputs...], name: <y>),
+
+  node((6,2), [outputs...], name: <u>),
+
+  edge((-0.3,3),(0.1,3), "-|>"),
+  node((0.4,3), [Task], stroke:none),
+  node((0,3.5), [], width:1em,height:1em),
+  node((0.4,3.5), [File ], stroke: none),
+
+  // Edges
+  bent-edge(<a>, <b>, [_Task 1_]),
+  bent-edge(<a>, <c>, [_Task 2_]),
+  bent-edge(<b>, <d>, [_Task 3_]),
+  bent-edge(<b>, <e>),
+  edge(<c>, <f>, "-|>", [_Task 4_]),
+  edge(<e>, <h>, "-|>"),
+  edge(<x>, <y>, "--|>", []),
+  edge(<w>, <u>, "--|>", []),
+  edge(<y>, <h>, "-|>"),
+  edge(<u>, <h>, "-|>"),
+)
+]
+
+

nextflow/.tutorial-answers.nf

@@ -0,0 +1,77 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl=2
+
+workflow {
+    reads_ch = Channel.fromPath("../input/*.fastq.gz")
+
+    kneaddata_out = kneaddata(reads_ch)
+    metaphlan_out = metaphlan(kneaddata_out[0], kneaddata_out[1])
+    humann_out = humann(metaphlan_out[0], metaphlan_out[1], metaphlan_out[2])
+}
+
+process kneaddata {
+    publishDir "kneaddata/"
+    input:
+    path file
+
+    output:
+    val sample
+    path "${sample}_kneaddata.fastq.gz"
+    path "${sample}_kneaddata*.fastq.gz"
+    path "${sample}_kneaddata.log"
+
+
+    script:
+    sample = file.name.replaceAll(".fastq.gz", "")
+
+    """
+    kneaddata --unpaired $file --output ./ --output-prefix ${sample}_kneaddata \
+        --reference-db ${params.kneaddata_db}
+
+    gzip *.fastq
+    """
+}
+
+process metaphlan {
+    publishDir "metaphlan/", mode: copy
+
+    input:
+    val sample
+    path knead_out
+
+    output:
+    val sample
+    path knead_out
+    path "${sample}_profile.tsv"
+
+    shell:
+
+    """
+    metaphlan $knead_out -o ${sample}_profile.tsv \
+        --input_type fastq
+    """
+}
+
+process humann {
+    publishDir "humann/", mode: copy
+
+    input:
+    val sample
+    path knead_out
+    path profile
+
+    output:
+    path "${sample}_genefamilies.tsv"
+    path "${sample}_pathabundance.tsv"
+    path "${sample}_pathcoverage.tsv"
+
+    shell:
+
+    """
+    humann --input $knead_out -o ./ --taxonomic-profile $profile \
+       --remove-temp-output --search-mode uniref90 \
+       --output-basename $sample
+    """
+}
+