Skip to content

Commit fc0e6d7

Browse files
committed
update structs
1 parent 589de09 commit fc0e6d7

File tree

2 files changed

+82
-72
lines changed

2 files changed

+82
-72
lines changed

src/utils/data_ingest.rs

Lines changed: 80 additions & 72 deletions
Original file line numberDiff line numberDiff line change
@@ -14,23 +14,23 @@ use std::{
1414
#[derive(Serialize, Deserialize, Debug, Clone)]
1515
pub struct CoverageData {
1616
#[serde(rename = "Sample")]
17-
sample_id: Option<String>,
17+
pub sample_id: Option<String>,
1818
#[serde(rename = "Reference_Name")]
19-
reference_name: String,
19+
pub reference_name: String,
2020
#[serde(rename = "Position")]
21-
position: String,
21+
pub position: String,
2222
#[serde(rename = "Coverage Depth")]
23-
coverage_depth: String,
23+
pub coverage_depth: String,
2424
#[serde(rename = "Consensus")]
25-
consensus: String,
25+
pub consensus: String,
2626
#[serde(rename = "Deletions")]
27-
deletions: String,
27+
pub deletions: String,
2828
#[serde(rename = "Ambiguous")]
29-
ambiguous: String,
29+
pub ambiguous: String,
3030
#[serde(rename = "Consensus_Count")]
31-
consensus_count: String,
31+
pub consensus_count: String,
3232
#[serde(rename = "Consensus_Average_Quality")]
33-
consensus_avg_quality: String,
33+
pub consensus_avg_quality: String,
3434
#[serde(rename = "Run_ID")]
3535
pub run_id: Option<String>,
3636
#[serde(rename = "Instrument")]
@@ -68,62 +68,70 @@ pub struct ProcessedRecord {
6868
}
6969

7070
/// Alleles struct
71-
#[derive(Serialize, Deserialize, Debug)]
71+
#[derive(Serialize, Deserialize, Debug, Clone)]
7272
pub struct AllelesData {
7373
#[serde(rename = "Reference_Name")]
74-
reference_name: String,
74+
pub reference_name: String,
7575
#[serde(rename = "HMM_Position")]
76-
reference_position: Option<String>,
76+
pub reference_position: Option<String>,
7777
#[serde(rename = "Position")]
78-
position: String,
78+
pub position: String,
7979
#[serde(rename = "Total")]
80-
total: String,
80+
pub total: String,
8181
#[serde(rename = "Consensus_Allele")]
82-
consensus_allele: String,
82+
pub consensus_allele: String,
8383
#[serde(rename = "Minority_Allele")]
84-
minority_allele: String,
84+
pub minority_allele: String,
8585
#[serde(rename = "Consensus_Count")]
86-
consensus_count: String,
86+
pub consensus_count: String,
8787
#[serde(rename = "Minority_Count")]
88-
minority_count: String,
88+
pub minority_count: String,
8989
#[serde(rename = "Minority_Frequency")]
90-
minority_frequency: String,
91-
sample_id: Option<String>,
90+
pub minority_frequency: String,
91+
pub sample_id: Option<String>,
92+
#[serde(rename = "Run_ID")]
93+
pub run_id: Option<String>,
94+
#[serde(rename = "Instrument")]
95+
pub instrument: Option<String>,
9296
}
9397

9498
/// Indel struct
95-
#[derive(Serialize, Deserialize, Debug)]
99+
#[derive(Serialize, Deserialize, Debug, Clone)]
96100
pub struct IndelsData {
97101
#[serde(rename = "Sample")]
98-
sample_id: Option<String>,
102+
pub sample_id: Option<String>,
99103
#[serde(rename = "Reference_Name")]
100-
reference_name: String,
104+
pub reference_name: String,
101105
#[serde(rename = "HMM_Position")]
102-
reference_upstream_position: Option<String>,
106+
pub reference_upstream_position: Option<String>,
103107
#[serde(rename = "Upstream_Position")]
104-
sample_upstream_position: Option<String>,
108+
pub sample_upstream_position: Option<String>,
105109
#[serde(rename = "Insert")]
106-
insert: Option<String>,
110+
pub insert: Option<String>,
107111
#[serde(rename = "Length")]
108-
length: Option<String>,
112+
pub length: Option<String>,
109113
#[serde(rename = "Context")]
110-
context: String,
114+
pub context: String,
111115
#[serde(rename = "Called")]
112-
called: String,
116+
pub called: String,
113117
#[serde(rename = "Count")]
114-
count: String,
118+
pub count: String,
115119
#[serde(rename = "Total")]
116-
total: String,
120+
pub total: String,
117121
#[serde(rename = "Frequency")]
118-
frequency: String,
122+
pub frequency: String,
119123
#[serde(rename = "Average_Quality")]
120-
average_quality: Option<String>,
124+
pub average_quality: Option<String>,
121125
#[serde(rename = "ConfidenceNotMacErr")]
122-
confidence_not_mac_err: Option<String>,
126+
pub confidence_not_mac_err: Option<String>,
123127
#[serde(rename = "PairedUB")]
124-
paired_ub: String,
128+
pub paired_ub: String,
125129
#[serde(rename = "QualityUB")]
126-
quality_ub: Option<String>,
130+
pub quality_ub: Option<String>,
131+
#[serde(rename = "Run_ID")]
132+
pub run_id: Option<String>,
133+
#[serde(rename = "Instrument")]
134+
pub instrument: Option<String>,
127135
}
128136

129137
#[derive(Debug)]
@@ -137,87 +145,87 @@ pub struct SeqData {
137145
#[derive(Serialize, Deserialize, Debug)]
138146
pub struct InsertionData {
139147
#[serde(rename = "ID")]
140-
sample_id: Option<String>,
148+
pub sample_id: Option<String>,
141149
#[serde(rename = "C_type")]
142-
ctype: Option<String>,
150+
pub ctype: Option<String>,
143151
#[serde(rename = "Ref_ID")]
144-
reference: String,
152+
pub reference: String,
145153
#[serde(rename = "Protein")]
146-
protein: String,
154+
pub protein: String,
147155
#[serde(rename = "Upstream_aa")]
148-
upstream_aa_position: String,
156+
pub upstream_aa_position: String,
149157
#[serde(rename = "Inserted_nucleotides")]
150-
inserted_nucleotides: String,
158+
pub inserted_nucleotides: String,
151159
#[serde(rename = "Inserted_residues")]
152-
inserted_residues: String,
160+
pub inserted_residues: String,
153161
#[serde(rename = "Upstream_nt")]
154-
upstream_nt: String,
162+
pub upstream_nt: String,
155163
#[serde(rename = "Codon_shift")]
156-
in_frame: String,
164+
pub in_frame: String,
157165
}
158166

159167
/// Deletions Data
160168
#[derive(Serialize, Deserialize, Debug)]
161169
pub struct DeletionsData {
162170
#[serde(rename = "ID")]
163-
sample_id: Option<String>,
171+
pub sample_id: Option<String>,
164172
#[serde(rename = "C_type")]
165-
ctype: Option<String>,
173+
pub ctype: Option<String>,
166174
#[serde(rename = "Ref_ID")]
167-
reference: String,
175+
pub reference: String,
168176
#[serde(rename = "Protein")]
169-
protein: String,
177+
pub protein: String,
170178
#[serde(rename = "VH")]
171-
vh: Option<String>,
179+
pub vh: Option<String>,
172180
#[serde(rename = "Del_AA_start")]
173-
del_start_aa_position: Option<String>,
181+
pub del_start_aa_position: Option<String>,
174182
#[serde(rename = "Del_AA_end")]
175-
del_end_aa_position: Option<String>,
183+
pub del_end_aa_position: Option<String>,
176184
#[serde(rename = "Del_AA_len")]
177-
del_aa_length: String,
185+
pub del_aa_length: String,
178186
#[serde(rename = "In_frame")]
179-
in_frame: String,
187+
pub in_frame: String,
180188
#[serde(rename = "CDS_ID")]
181-
cds_id: Option<String>,
189+
pub cds_id: Option<String>,
182190
#[serde(rename = "Del_CDS_start")]
183-
del_start_cds_position: String,
191+
pub del_start_cds_position: String,
184192
#[serde(rename = "Del_CDS_end")]
185-
del_end_cds_position: String,
193+
pub del_end_cds_position: String,
186194
#[serde(rename = "Del_CDS_len")]
187-
del_cds_length: Option<String>,
195+
pub del_cds_length: Option<String>,
188196
}
189197

190198
/// Sequence Data
191199
#[derive(Serialize, Deserialize, Debug)]
192200
pub struct DaisSeqData {
193201
#[serde(rename = "ID")]
194-
sample_id: Option<String>,
202+
pub sample_id: Option<String>,
195203
#[serde(rename = "C_type")]
196-
ctype: Option<String>,
204+
pub ctype: Option<String>,
197205
#[serde(rename = "Ref_ID")]
198-
reference: String,
206+
pub reference: String,
199207
#[serde(rename = "Protein")]
200-
protein: String,
208+
pub protein: String,
201209
#[serde(rename = "VH")]
202-
vh: Option<String>,
210+
pub vh: Option<String>,
203211
#[serde(rename = "AA_seq")]
204-
aa_seq: String,
212+
pub aa_seq: String,
205213
#[serde(rename = "AA_aln")]
206-
aa_aln: String,
214+
pub aa_aln: String,
207215
#[serde(rename = "CDS_ID")]
208-
cds_id: Option<String>,
216+
pub cds_id: Option<String>,
209217
#[serde(rename = "Insertion")]
210-
insertion: String,
218+
pub insertion: String,
211219
#[serde(rename = "Shift_Insert")]
212-
insertions_shift_frame: String,
220+
pub insertions_shift_frame: String,
213221
#[serde(rename = "CDS_seq")]
214-
cds_sequence: String,
222+
pub cds_sequence: String,
215223
#[serde(rename = "CDS_aln")]
216-
aligned_cds_sequence: String,
224+
pub aligned_cds_sequence: String,
217225
#[serde(rename = "Query_nt_coordinates")]
218-
reference_nt_positions: String,
226+
pub reference_nt_positions: String,
219227
#[serde(rename = "CDS_nt_coordinates")]
220-
sample_nt_positions: String,
228+
pub sample_nt_positions: String,
221229
}
222230

223231
/////////////// Imp for the process_txt_with_sample_function ///////////////

src/utils/writing_outputs.rs

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -9,6 +9,8 @@ use serde_json::{Value, json};
99
use std::sync::Arc;
1010
use std::{error::Error, fs::File};
1111

12+
use super::data_ingest::CoverageData;
13+
1214
/////////////// Functions to write to json and csv files ///////////////
1315
/// Function to serialize a vector of structs into split-oriented JSON with precision and indexing
1416
pub fn write_structs_to_split_json_file<T: Serialize>(

0 commit comments

Comments
 (0)