Merge pull request #20 from RasmussenLab/dbds_update

Dbds update

Author: madscort

bin/build_matrix.py

@@ -35,9 +35,6 @@
 from pathlib import Path
 from typing import Dict, List, Tuple, Optional
 
-def natural_key(s: str):
-    return [int(t) if t.isdigit() else t.lower() for t in re.findall(r"\d+|\D+", s)]
-
 def read_pools(pools_path: Path) -> Tuple[List[str], List[str]]:
     rows, cols = [], []
     with pools_path.open() as fh:
@@ -108,8 +105,8 @@ def build_context(rows: List[str],
                   decode: Dict[Tuple[str,str], str],
                   pad_width: Optional[int]) -> dict:
     # Force deterministic ordering
-    rows_ord = sorted(rows, key=natural_key)
-    cols_ord = sorted(cols, key=natural_key)
+    rows_ord = sorted(rows, key=str.lower)
+    cols_ord = sorted(cols, key=str.lower)
 
     # Indices and labels
     row_index = {pid: i for i, pid in enumerate(rows_ord)}

bin/report.Rmd

@@ -60,7 +60,8 @@ output_path <- here::here('reports')
 pinvars <- 
   read_tsv(pinpoints_path,
            col_types = cols(.default = col_character())) %>% 
-  type_convert(guess_integer = TRUE)
+  type_convert(guess_integer = TRUE) %>% 
+  mutate(row_label = as.integer(row_label))
 
 poolvars <- 
   read_tsv(all_variants_path,
@@ -71,35 +72,22 @@ poolvars <-
 ```{r eval=have(params$rescue)}
 rescue <- 
   read_tsv(rescue_path) %>% 
+  mutate(row_label = as.integer(row_label)) %>% 
   select(-sample_id)
 
 pinvars <- 
   pinvars %>% 
-  full_join(rescue, by=c('uvarid','row_id','column_id','varid','row_label','column_label','sample_alias')) 
+  full_join(rescue, by=c('uvarid','row_id','column_id','row_label','column_label','varid','sample_alias')) %>% 
+  mutate(is_pool_pin = if_else(is.na(is_pool_pin),0,is_pool_pin))
 ```
 
 ```{r eval=have(params$annotations)}
 
-annotations <- 
-  read_tsv(annotations_path,
-           na = '-', col_types = cols(.default = col_character())) %>% 
-  type_convert(guess_integer = TRUE)
-
-pinvars <- 
-  pinvars %>% 
-  left_join(annotations, by='varid')
-
-poolvars <- 
-  poolvars %>% 
-  left_join(annotations, by='varid')
-```
-
-
-```{r eval=F}
 annotations <- 
   read_tsv(annotations_path,
            na = '-', col_types = cols(.default = col_character())) %>% 
   type_convert(guess_integer = TRUE) %>% 
+  mutate(is_lof = if_else(LoF == 'HC',1,0)) %>% 
   mutate(clinvar_stars = case_when(
     str_detect(ClinVar_CLNREVSTAT, "practice_guideline") ~ 4,
     str_detect(ClinVar_CLNREVSTAT, "reviewed_by_expert_panel") ~ 3,
@@ -111,23 +99,13 @@ annotations <-
   mutate(is_p = if_else(str_detect(ClinVar_CLNSIG, 'Pathogenic') | str_detect(ClinVar_CLNSIG, "Likely_pathogenic"), 1,0)) %>% 
   mutate(is_lofp = if_else(is_lof == 1 | is_p == 1,1,0))
 
-rescue <- 
-  read_tsv(rescue_path) %>% 
-  select(-sample_id)
-
 pinvars <- 
-  read_tsv(pinpoints_path,
-           col_types = cols(.default = col_character())) %>% 
-  type_convert(guess_integer = TRUE) %>% 
-  left_join(rescue, by=c('uvarid','row_id','column_id','varid')) %>% 
+  pinvars %>% 
   left_join(annotations, by='varid')
 
 poolvars <- 
-  read_tsv(all_variants_path,
-           col_types = cols(.default = col_character())) %>% 
-  type_convert(guess_integer = TRUE) %>% 
+  poolvars %>% 
   left_join(annotations, by='varid')
-
 ```
 
 ```{r}
@@ -166,8 +144,10 @@ pinvars <-
 ```{r}
 plot_data <-
   pinvars %>% 
+  filter(is_pool_pin == 1) %>% 
   group_by(sample_id,row_factor, column_factor) %>% 
   summarise(private_variants = n())
+
 plot_data %>% 
   ggplot(aes(x = column_factor, y = row_factor, fill = private_variants)) +
   geom_point(shape = 22, size = 14, color='black') +
@@ -190,8 +170,8 @@ plot_data %>%
 ```{r eval=have(params$annotations)}
 plot_data <-
   pinvars %>% 
-  mutate(row_factor = fct_rev(factor(row_label, levels = 1:2, ordered = TRUE)),
-         column_factor = factor(column_label, levels = LETTERS[1:2], ordered = TRUE)) %>% 
+  filter(is_pool_pin == 1) %>% 
+  filter(is_lofp == 1) %>% 
   group_by(sample_id,row_factor, column_factor) %>% 
   summarise(private_variants = n())
 
@@ -214,11 +194,15 @@ plot_data %>%
   ggtitle('pLoF/p only')
 ```
 
+
+
+
+
 ```{r eval=have(params$annotations) & have(params$rescue)}
 plot_data <-
   pinvars %>% 
-  mutate(row_factor = fct_rev(factor(row_label, levels = 1:2, ordered = TRUE)),
-         column_factor = factor(column_label, levels = LETTERS[1:2], ordered = TRUE)) %>% 
+  filter((probability >= 0.5 & is_single) | is_pool_pin == 1) %>% 
+  filter(is_lofp == 1) %>% 
   group_by(sample_id,row_factor, column_factor) %>% 
   summarise(private_variants = n())
 

conf/container.config

@@ -34,7 +34,7 @@ params {
             pinpy = "madscort/dswf_pinpy:1.0"
             filter_variants = "madscort/dswf_filter:1.0"
             vep = "madscort/vep:111.0"
-            marbl = "madscort/marbl:0.4.0"
+            marbl = "madscort/marbl:0.6.0"
         }
         singularity {
             gatk = "https://depot.galaxyproject.org-singularity-gatk4-4.5.0.0--py36hdfd78af_0"
@@ -67,7 +67,7 @@ params {
             pinpy = "madscort/dswf_pinpy:1.0"
             filter_variants = "madscort/dswf_filter:1.0"
             vep = "madscort/vep:111.0"
-            marbl = "madscort/marbl:0.4.0"
+            marbl = "madscort/marbl:0.6.0"
         }
     }
 }

conf/ngc.config

@@ -107,12 +107,6 @@ profiles {
                 time = 2.hour
             }
 
-            withName: DUPLICATE_METRICS {
-                // module = ['tools','oracle_jdk/21.0.2','gatk/4.6.0.0']
-                cpus = 4
-                memory = 16.GB
-                time = 2.hour
-            }
 
             withName: GROUP_UMI {
                 // module = ['tools','java/1.8.0','fgbio/1.5.1']

envs/pinpy/build

@@ -1 +1 @@
-docker buildx build --platform linux/amd64,linux/arm64,windows/amd64 -t madscort/dswf_pinpy:1.0 --push .
+docker buildx build --platform linux/amd64,linux/arm64 -t madscort/dswf_pinpy:1.0 --push .

envs/pinpy/environment.yaml

@@ -4,5 +4,6 @@ channels:
   - bioconda
   - defaults
 dependencies:
-  - bcftools=1.19
-  - python=3.9
+  - bcftools=1.22
+  - python=3.9
+  - samtools=1.22

envs/rescue/build

@@ -1 +1 @@
-docker buildx build --platform linux/amd64,linux/arm64 -t madscort/marbl:0.4.0 --push .
+docker buildx build --platform linux/amd64,linux/arm64 -t madscort/marbl:0.6.0 --push .

envs/rescue/environment.yaml

@@ -6,4 +6,4 @@ channels:
 dependencies:
   - python=3.11
   - pip:
-    - marbl-pool==0.4.0
+    - marbl-pool==0.6.0

next.pbs

@@ -14,7 +14,7 @@
 #    myqsub [-F "optional arguments to Nexflow"] next.pbs
 #
 
-#PBS -l nodes=1:ppn=40:thinnode
+#PBS -l nodes=1:ppn=40
 #PBS -l walltime=24:00:00
 #PBS -l mem=160gb
 #PBS -j oe