Add option to deactivate the BLAST default DUST filter for low comple…

…xity sequences

.tests/config/config.yaml

@@ -26,6 +26,7 @@ taxdb: data/miniblast
 taxid_filter: 7742
 blocklist: extinct
 seq_blocklist: None
+blast_filter_low_complexity: True
 blast_evalue: 1e-10
 blast_identity: 97
 blast_qcov: 100

.tests/config/config_foodme_paramspace.yaml

@@ -28,6 +28,7 @@ taxdb: ../../../data/miniblast
 taxid_filter: 7742
 blocklist: extinct
 seq_blocklist: None
+blast_filter_low_complexity: True
 blast_evalue: 1e-10
 blast_identity: 97
 blast_qcov: 100

.tests/config/config_otu.yaml

@@ -26,6 +26,7 @@ taxdb: data/miniblast
 taxid_filter: 7742
 blocklist: extinct
 seq_blocklist: None
+blast_filter_low_complexity: True
 blast_evalue: 1e-10
 blast_identity: 97
 blast_qcov: 100

CHANGELOG.md

@@ -1,3 +1,24 @@
+### 1.7.0
+
+#### Breaking changes
+
+Older configuration files are not compatible anymore. Update the files by adding the following line
+should yield the same results as before:
+
+```{yaml}
+blast_filter_low_complexity: True
+```
+
+#### New features
+
+It is now possible to desactivate the defualt low-complexity filter of the BLAST search.
+This can be advantageous if you expect your barcode to contain low-complexity sequences which could 
+prevent getting any match at all.
+This behaviour can be activated/deactivated by changing the `blast_filter_low_complexity` from `True` to `False`.
+
+The default behaviour (`False`) uses the default 'DUST' filter of the blast tool:
+`-dust 20 64 1 -soft_masking true`.
+
 ### 1.6.6
 
 #### Fixes

VERSION

@@ -1 +1 @@
-1.6.6
+1.7.0

config/config.yaml

@@ -73,6 +73,8 @@ blocklist: extinct
 # Exclude specific sequence accessions from the results.
 # None or path to a user provided list of accessions
 seq_blocklist: None
+# BLAST low-complexity filter 
+blast_filter_low_complexity: True
 # E-value threshold for blast results
 blast_evalue: 1e-10
 # Minimal identity between the hit and query for blast results (in percent)

config/config_16Smeat.yaml

@@ -90,6 +90,8 @@ blocklist: extinct
 # Sequence Accessions blocklist, prevents specific accessions to show up in the results.
 # None or path to a user provided list of accessions
 seq_blocklist: None
+# BLAST low-complexity filter 
+blast_filter_low_complexity: True
 # E-value threshold for blast results
 blast_evalue: 1e-20
 # Minimal identity between the hit and query for blast results (in percent)

docs/userguide/configuration.md

@@ -84,6 +84,7 @@ This will create a file called `samples.tsv` in the `raw_data` folder.
 | `taxid_filter`            | Taxonomic identifier      | Node under which to perform the BLAST search. <br>Equivalent to pruning the taxonomy above <br>this node. Use the Root Node number to keep <br>the entire taxonomy |
 | `blocklist`               | `extinct` or custom path  | Path to a list of taxonomic identifier to exclude <br>from the BLAST search |
 | `seq_blocklist`           | `None` or custom path | Path to a list of sequence accessions (e.g. `NC_0016400`) <br>to exclude from the results |
+| `blast_filter_low_complexity` | True/False | Wether to mask low-complexity regions in the BLAST search. On by default, deactivate if you expect barcode sequences with low complexity. | 
 | `blast_evalue`                 | Number (scientific)       | Minimal E-value threshold for the BLAST search |
 | `blast_identity`           | Number [0, 100]           | Minimal identity (in percent) between the query and <br>hit sequence for the BLAST search |
 | `blast_qcov`                    | Number [0, 100]           | Percent of the query to be covered by the hit <br>sequence for the BLAST search |

workflow/rules/blast.smk

@@ -127,6 +127,7 @@ rule blast_otus:
         e_value=config["blast_evalue"],
         perc_identity=config["blast_identity"],
         qcov=config["blast_qcov"],
+        low_complexity=get_low_complexity_filter_params,
     threads: config["threads_sample"]
     message:
         "[{wildcards.sample}][assignement] BLASTing clusters against local database"
@@ -148,7 +149,7 @@ rule blast_otus:
         blastn -db {params.blast_DB} \
             -query {input.query} \
             -out {output.report} \
-            -task 'megablast' \
+            -task 'megablast' {params.low_complexity} \
             -evalue {params.e_value} \
             -perc_identity {params.perc_identity} \
             -qcov_hsp_perc {params.qcov} $masking \

workflow/rules/common.smk

@@ -51,3 +51,11 @@ def get_acc_blocklist(wildcards):
         return f"{wildcards.sample}/taxonomy/{wildcards.sample}_blast_report.tsv"
     else:
         return f"{wildcards.sample}/taxonomy/{wildcards.sample}_blast_report_prefiltered.tsv"
+
+
+def get_low_complexity_filter_params(wildcards):
+    # Filter is on by default in blastn
+    if config["blast_filter_low_complexity"]:
+        return ""
+    else:
+        return "-dust no -soft_masking false"

workflow/rules/common_benchmark.smk

@@ -62,3 +62,11 @@ def get_acc_blocklist(wildcards):
         return f"{wildcards.sample}/taxonomy/{wildcards.sample}_blast_report.tsv"
     else:
         return f"{wildcards.sample}/taxonomy/{wildcards.sample}_blast_report_prefiltered.tsv"
+
+
+def get_low_complexity_filter_params(wildcards):
+    # Filter is on by default in blastn
+    if config["blast_filter_low_complexity"]:
+        return ""
+    else:
+        return "-dust no -soft_masking false"