added STRT-seq

examples/STRT-seq-2i/spec.yaml

@@ -0,0 +1,126 @@
+assay:
+  name: STRT-seq-2i
+  doi: https://doi.org/10.1038/s41598-017-16546-4
+  description: STRT-seq performed on microwells with dual-index 5 prime
+  modalities: [RNA]
+  lib_struct: https://teichlab.github.io/scg_lib_structs/methods_html/STRT-seq_family.html
+  assay_spec:
+    RNA:
+      seq_type: joined
+      seq: 
+      min_len: 
+      max_len: 
+      onlist: 
+      join:
+        how: union
+        order: [solexa_p1, link_1, well_bc, link_2, umi, triple_G, cDNA, ME, link_3, subarray_bc, solexa_p2]
+        regions:
+          solexa_p1: 
+            seq_type: fixed
+            seq: AATGATACGGCGACCACCGA
+            min_len: 20
+            max_len: 21
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          link_1:
+            seq_type: fixed
+            seq: GATCTACAC
+            min_len: 9
+            max_len: 10
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          well_bc:
+            seq_type: random
+            seq: NNNNN
+            min_len: 5
+            max_len: 6
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          link_2:
+            seq_type: fixed
+            seq: CTACACGACGCTCTTCCGATCT
+            min_len: 22
+            max_len: 23
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          umi:
+            seq_type: random
+            seq: NNNNNN
+            min_len: 6
+            max_len: 7
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          triple_G:
+            seq_type: fixed
+            seq: GGG
+            min_len: 3
+            max_len: 4
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          cDNA:
+            seq_type: random
+            seq: 
+            min_len: 1
+            max_len: 99
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          ME:
+            seq_type: fixed
+            seq: CTGTCTCTTATACACATCT
+            min_len: 19
+            max_len: 20
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          link_3:
+            seq_type: fixed
+            seq: GACGC
+            min_len: 5
+            max_len: 6
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          subarray_bc:
+            seq_type: random
+            seq: NNNNNNNN
+            min_len: 8
+            max_len: 9
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          solexa_p2:
+            seq_type: fixed
+            seq: TCGTATGCCGTCTTCTGCTTG
+            min_len: 21
+            max_len: 22
+            onlist: 
+            join: 
+              regions:
+              how: 

examples/STRT-seq-C1/spec.yaml

@@ -0,0 +1,106 @@
+assay:
+  name: STRT-seq-C1
+  doi: https://doi.org/10.1007/978-1-4939-9240-9_9
+  description: STRT-seq performed on microfluidics Fluidigm C1
+  modalities: [RNA]
+  lib_struct: https://teichlab.github.io/scg_lib_structs/methods_html/STRT-seq_family.html
+  assay_spec:
+    RNA:
+      seq_type: joined
+      seq: 
+      min_len: 
+      max_len: 
+      onlist: 
+      join:
+        how: union
+        order: [solexa_p1, link_1, umi, triple_G, cDNA, ME, link_2, cell_bc, solexa_p2]
+        regions:
+          solexa_p1: 
+            seq_type: fixed
+            seq: AATGATACGGCGACCACCGA
+            min_len: 20
+            max_len: 21
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          link_1:
+            seq_type: fixed
+            seq: T
+            min_len: 1
+            max_len: 2
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          umi:
+            seq_type: random
+            seq: NNNNN
+            min_len: 5
+            max_len: 6
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          triple_G:
+            seq_type: fixed
+            seq: GGG
+            min_len: 3
+            max_len: 4
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          cDNA:
+            seq_type: random
+            seq: 
+            min_len: 1
+            max_len: 99
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          ME:
+            seq_type: fixed
+            seq: CTGTCTCTTATACACATCT
+            min_len: 19
+            max_len: 20
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          link_2:
+            seq_type: fixed
+            seq: GACGC
+            min_len: 5
+            max_len: 6
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          cell_bc:
+            seq_type: random
+            seq: NNNNNNNN
+            min_len: 8
+            max_len: 9
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          solexa_p2:
+            seq_type: fixed
+            seq: TCGTATGCCGTCTTCTGCTTG
+            min_len: 21
+            max_len: 22
+            onlist: 
+            join: 
+              regions:
+              how: 

examples/STRT-seq/spec.yaml

@@ -0,0 +1,106 @@
+assay:
+  name: STRT-seq
+  doi: https://doi.org/10.1038/nprot.2012.022
+  description: single-cell tagged reverse transcripton
+  modalities: [RNA]
+  lib_struct: https://teichlab.github.io/scg_lib_structs/methods_html/STRT-seq_family.html
+  assay_spec:
+    RNA:
+      seq_type: joined
+      seq: 
+      min_len: 
+      max_len: 
+      onlist: 
+      join:
+        how: union
+        order: [solexa_p1, link_1, ispcr, link_2, cell_bc, triple_G, cDNA, link_3, solexa_p2]
+        regions:
+          solexa_p1: 
+            seq_type: fixed
+            seq: AATGATACGGCGACCACCGA
+            min_len: 20
+            max_len: 21
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          link_1:
+            seq_type: fixed
+            seq: GATCT
+            min_len: 5
+            max_len: 6
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          ispcr:
+            seq_type: fixed
+            seq: AAGCAGTGGTATCAACGCAGAGT
+            min_len: 23
+            max_len: 24
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          link_2:
+            seq_type: fixed
+            seq: GCAGTGCT
+            min_len: 8
+            max_len: 9
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          cell_bc:
+            seq_type: random
+            seq: NNNNNN
+            min_len: 6
+            max_len: 7
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          triple_G:
+            seq_type: fixed
+            seq: GGG
+            min_len: 3
+            max_len: 4
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          cDNA:
+            seq_type: random
+            seq: 
+            min_len: 1
+            max_len: 99
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          link_3:
+            seq_type: fixed
+            seq: AGATCGGAAGAGC
+            min_len: 13
+            max_len: 14
+            onlist: 
+            join: 
+              regions:
+              how: 
+
+          solexa_p2:
+            seq_type: fixed
+            seq: TCGTATGCCGTCTTCTGCTTG
+            min_len: 21
+            max_len: 22
+            onlist: 
+            join: 
+              regions:
+              how: