@sjcrh/proteinpaint-rust 2.116.0 → 2.119.0-0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (5) hide show
  1. package/Cargo.toml +2 -1
  2. package/package.json +2 -2
  3. package/src/DEanalysis.rs +51 -44
  4. package/src/readHDF5.rs +809 -240
  5. package/src/topGeneByExpressionVariance.rs +144 -144
package/src/topGeneByExpressionVariance.rs CHANGED
@@ -64,15 +64,15 @@ fn input_data_hdf5(
64
64
  let file = match File::open(filename) {
65
65
  Ok(f) => f,
66
66
  Err(err) => {
67
- eprintln!("Failed to open HDF5 file: {}", err);
68
- println!(
69
- "{}",
70
- serde_json::json!({
71
- "status": "error",
72
- "message": format!("Failed to open HDF5 file: {}", err),
73
- "file_path": filename
74
- })
75
- );
67
+ // eprintln!("Failed to open HDF5 file: {}", err);
68
+ // println!(
69
+ // "{}",
70
+ // serde_json::json!({
71
+ // "status": "error",
72
+ // "message": format!("Failed to open HDF5 file: {}", err),
73
+ // "file_path": filename
74
+ // })
75
+ // );
76
76
  return Err(hdf5::Error::Internal(format!(
77
77
  "Failed to open HDF5 file: {}",
78
78
  err
@@ -81,20 +81,20 @@ fn input_data_hdf5(
81
81
  };
82
82
 
83
83
  // Read gene symbols dataset
84
- let genes_dataset = match file.dataset("gene_symbols") {
84
+ let genes_dataset = match file.dataset("gene_names") {
85
85
  Ok(ds) => ds,
86
86
  Err(err) => {
87
- eprintln!("Failed to open gene_symbols dataset: {}", err);
88
- println!(
89
- "{}",
90
- serde_json::json!({
91
- "status": "error",
92
- "message": format!("Failed to open gene_symbols dataset: {}", err),
93
- "file_path": filename
94
- })
95
- );
87
+ // eprintln!("Failed to open gene_names dataset: {}", err);
88
+ // println!(
89
+ // "{}",
90
+ // serde_json::json!({
91
+ // "status": "error",
92
+ // "message": format!("Failed to open gene_names dataset: {}", err),
93
+ // "file_path": filename
94
+ // })
95
+ // );
96
96
  return Err(hdf5::Error::Internal(format!(
97
- "Failed to open gene_symbols dataset: {}",
97
+ "Failed to open gene_names dataset: {}",
98
98
  err
99
99
  )));
100
100
  }
@@ -104,15 +104,15 @@ fn input_data_hdf5(
104
104
  let genes_varlen = match genes_dataset.read_1d::<VarLenAscii>() {
105
105
  Ok(g) => g,
106
106
  Err(err) => {
107
- eprintln!("Failed to read gene symbols: {}", err);
108
- println!(
109
- "{}",
110
- serde_json::json!({
111
- "status": "error",
112
- "message": format!("Failed to read gene symbols: {}", err),
113
- "file_path": filename
114
- })
115
- );
107
+ // eprintln!("Failed to read gene symbols: {}", err);
108
+ // println!(
109
+ // "{}",
110
+ // serde_json::json!({
111
+ // "status": "error",
112
+ // "message": format!("Failed to read gene symbols: {}", err),
113
+ // "file_path": filename
114
+ // })
115
+ // );
116
116
  return Err(hdf5::Error::Internal(format!(
117
117
  "Failed to read gene symbols: {}",
118
118
  err
@@ -121,15 +121,15 @@ fn input_data_hdf5(
121
121
  };
122
122
 
123
123
  // Convert to Vec<String> for easier handling
124
- let gene_symbols: Vec<String> = genes_varlen.iter().map(|g| g.to_string()).collect();
125
- let num_genes = gene_symbols.len();
124
+ let gene_names: Vec<String> = genes_varlen.iter().map(|g| g.to_string()).collect();
125
+ let num_genes = gene_names.len();
126
126
  // eprintln!("Found {} gene symbols", num_genes);
127
127
 
128
128
  // Read sample names
129
129
  let samples_dataset = match file.dataset("samples") {
130
130
  Ok(ds) => ds,
131
131
  Err(err) => {
132
- eprintln!("Failed to open samples dataset: {}", err);
132
+ // eprintln!("Failed to open samples dataset: {}", err);
133
133
  println!(
134
134
  "{}",
135
135
  serde_json::json!({
@@ -149,7 +149,7 @@ fn input_data_hdf5(
149
149
  let samples_varlen = match samples_dataset.read_1d::<VarLenAscii>() {
150
150
  Ok(s) => s,
151
151
  Err(err) => {
152
- eprintln!("Failed to read sample names: {}", err);
152
+ // eprintln!("Failed to read sample names: {}", err);
153
153
  println!(
154
154
  "{}",
155
155
  serde_json::json!({
@@ -175,16 +175,16 @@ fn input_data_hdf5(
175
175
  if let Some(index) = all_samples.iter().position(|s| s == sample) {
176
176
  column_indices.push(index);
177
177
  } else {
178
- eprintln!("Sample {} not found in the dataset", sample);
179
- println!(
180
- "{}",
181
- serde_json::json!({
182
- "status": "error",
183
- "message": format!("Sample '{}' not found in the dataset", sample),
184
- "file_path": filename,
185
- "available_samples": all_samples
186
- })
187
- );
178
+ // eprintln!("Sample {} not found in the dataset", sample);
179
+ // println!(
180
+ // "{}",
181
+ // serde_json::json!({
182
+ // "status": "error",
183
+ // "message": format!("Sample '{}' not found in the dataset", sample),
184
+ // "file_path": filename,
185
+ // "available_samples": all_samples
186
+ // })
187
+ // );
188
188
  return Err(hdf5::Error::Internal(format!(
189
189
  "Sample '{}' not found in the dataset",
190
190
  sample
@@ -196,15 +196,15 @@ fn input_data_hdf5(
196
196
  let counts_dataset = match file.dataset("counts") {
197
197
  Ok(ds) => ds,
198
198
  Err(err) => {
199
- eprintln!("Failed to open counts dataset: {}", err);
200
- println!(
201
- "{}",
202
- serde_json::json!({
203
- "status": "error",
204
- "message": format!("Failed to open counts dataset: {}", err),
205
- "file_path": filename
206
- })
207
- );
199
+ // eprintln!("Failed to open counts dataset: {}", err);
200
+ // println!(
201
+ // "{}",
202
+ // serde_json::json!({
203
+ // "status": "error",
204
+ // "message": format!("Failed to open counts dataset: {}", err),
205
+ // "file_path": filename
206
+ // })
207
+ // );
208
208
  return Err(hdf5::Error::Internal(format!(
209
209
  "Failed to open counts dataset: {}",
210
210
  err
@@ -215,16 +215,16 @@ fn input_data_hdf5(
215
215
  // Get dataset dimensions for validation
216
216
  let dataset_shape = counts_dataset.shape();
217
217
  if dataset_shape.len() != 2 {
218
- eprintln!("Counts dataset does not have the expected 2D shape");
219
- println!(
220
- "{}",
221
- serde_json::json!({
222
- "status": "error",
223
- "message": "Expected a 2D dataset for counts",
224
- "file_path": filename,
225
- "actual_shape": dataset_shape
226
- })
227
- );
218
+ // eprintln!("Counts dataset does not have the expected 2D shape");
219
+ // println!(
220
+ // "{}",
221
+ // serde_json::json!({
222
+ // "status": "error",
223
+ // "message": "Expected a 2D dataset for counts",
224
+ // "file_path": filename,
225
+ // "actual_shape": dataset_shape
226
+ // })
227
+ // );
228
228
  return Err(hdf5::Error::Internal(
229
229
  "Expected a 2D dataset for counts".to_string(),
230
230
  ));
@@ -232,19 +232,19 @@ fn input_data_hdf5(
232
232
 
233
233
  // Check dimensions match expected values
234
234
  if dataset_shape[0] != num_genes {
235
- eprintln!(
236
- "Counts dataset first dimension ({}) doesn't match number of genes ({})",
237
- dataset_shape[0], num_genes
238
- );
239
- println!(
240
- "{}",
241
- serde_json::json!({
242
- "status": "error",
243
- "message": format!("Counts dataset first dimension ({}) doesn't match number of genes ({})",
244
- dataset_shape[0], num_genes),
245
- "file_path": filename
246
- })
247
- );
235
+ // eprintln!(
236
+ // "Counts dataset first dimension ({}) doesn't match number of genes ({})",
237
+ // dataset_shape[0], num_genes
238
+ // );
239
+ // println!(
240
+ // "{}",
241
+ // serde_json::json!({
242
+ // "status": "error",
243
+ // "message": format!("Counts dataset first dimension ({}) doesn't match number of genes ({})",
244
+ // dataset_shape[0], num_genes),
245
+ // "file_path": filename
246
+ // })
247
+ // );
248
248
  return Err(hdf5::Error::Internal(format!(
249
249
  "Counts dataset first dimension ({}) doesn't match number of genes ({})",
250
250
  dataset_shape[0], num_genes
@@ -252,20 +252,20 @@ fn input_data_hdf5(
252
252
  }
253
253
 
254
254
  if dataset_shape[1] != all_samples.len() {
255
- eprintln!(
256
- "Counts dataset second dimension ({}) doesn't match number of samples ({})",
257
- dataset_shape[1],
258
- all_samples.len()
259
- );
260
- println!(
261
- "{}",
262
- serde_json::json!({
263
- "status": "error",
264
- "message": format!("Counts dataset second dimension ({}) doesn't match number of samples ({})",
265
- dataset_shape[1], all_samples.len()),
266
- "file_path": filename
267
- })
268
- );
255
+ // eprintln!(
256
+ // "Counts dataset second dimension ({}) doesn't match number of samples ({})",
257
+ // dataset_shape[1],
258
+ // all_samples.len()
259
+ // );
260
+ // println!(
261
+ // "{}",
262
+ // serde_json::json!({
263
+ // "status": "error",
264
+ // "message": format!("Counts dataset second dimension ({}) doesn't match number of samples ({})",
265
+ // dataset_shape[1], all_samples.len()),
266
+ // "file_path": filename
267
+ // })
268
+ // );
269
269
  return Err(hdf5::Error::Internal(format!(
270
270
  "Counts dataset second dimension ({}) doesn't match number of samples ({})",
271
271
  dataset_shape[1],
@@ -277,15 +277,15 @@ fn input_data_hdf5(
277
277
  let all_counts = match counts_dataset.read::<f64, Dim<[usize; 2]>>() {
278
278
  Ok(data) => data,
279
279
  Err(err) => {
280
- eprintln!("Failed to read expression data: {}", err);
281
- println!(
282
- "{}",
283
- serde_json::json!({
284
- "status": "error",
285
- "message": format!("Failed to read expression data: {}", err),
286
- "file_path": filename
287
- })
288
- );
280
+ // eprintln!("Failed to read expression data: {}", err);
281
+ // println!(
282
+ // "{}",
283
+ // serde_json::json!({
284
+ // "status": "error",
285
+ // "message": format!("Failed to read expression data: {}", err),
286
+ // "file_path": filename
287
+ // })
288
+ // );
289
289
  return Err(hdf5::Error::Internal(format!(
290
290
  "Failed to read expression data: {}",
291
291
  err
@@ -316,7 +316,7 @@ fn input_data_hdf5(
316
316
  // dm.ncols()
317
317
  // );
318
318
 
319
- Ok((dm, gene_symbols))
319
+ Ok((dm, gene_names))
320
320
  }
321
321
 
322
322
  // The original input_data function for text files is kept as is
@@ -330,7 +330,7 @@ fn input_data(
330
330
  // Build the CSV reader and iterate over each record.
331
331
  let mut reader = BGZFReader::new(fs::File::open(filename).unwrap()).unwrap();
332
332
  let mut num_lines: usize = 0;
333
- let mut gene_symbols: Vec<String> = Vec::with_capacity(500);
333
+ let mut gene_names: Vec<String> = Vec::with_capacity(500);
334
334
 
335
335
  let mut buffer = String::new();
336
336
  reader.read_to_string(&mut buffer).unwrap();
@@ -358,7 +358,7 @@ fn input_data(
358
358
  } else {
359
359
  num_lines += 1;
360
360
  //println!("line2:{:?}", line2);
361
- gene_symbols.push(line2[3].to_string());
361
+ gene_names.push(line2[3].to_string());
362
362
  for i in &column_numbers {
363
363
  let field = line2[*i];
364
364
  let num = FromStr::from_str(field);
@@ -386,7 +386,7 @@ fn input_data(
386
386
 
387
387
  let dm = DMatrix::from_row_slice(num_lines, sample_list.len(), &input_vector);
388
388
  //println!("dm:{:?}", dm);
389
- (dm, gene_symbols)
389
+ (dm, gene_names)
390
390
  }
391
391
 
392
392
  #[allow(dead_code)]
@@ -398,7 +398,7 @@ struct GeneInfo {
398
398
 
399
399
  fn calculate_variance(
400
400
  input_matrix: Matrix<f64, Dyn, Dyn, VecStorage<f64, Dyn, Dyn>>,
401
- gene_symbols: Vec<String>,
401
+ gene_names: Vec<String>,
402
402
  mut min_sample_size: f64,
403
403
  filter_extreme_values: bool,
404
404
  rank_type: String,
@@ -478,12 +478,12 @@ fn calculate_variance(
478
478
  {
479
479
  gene_infos.push(GeneInfo {
480
480
  rank_type: gene_counts.variance(),
481
- gene_symbol: gene_symbols[row].clone(),
481
+ gene_symbol: gene_names[row].clone(),
482
482
  });
483
483
  } else if filter_extreme_values == false {
484
484
  gene_infos.push(GeneInfo {
485
485
  rank_type: gene_counts.variance(),
486
- gene_symbol: gene_symbols[row].clone(),
486
+ gene_symbol: gene_names[row].clone(),
487
487
  });
488
488
  }
489
489
  } else {
@@ -496,12 +496,12 @@ fn calculate_variance(
496
496
  {
497
497
  gene_infos.push(GeneInfo {
498
498
  rank_type: gene_counts_data.interquartile_range(),
499
- gene_symbol: gene_symbols[row].clone(),
499
+ gene_symbol: gene_names[row].clone(),
500
500
  });
501
501
  } else if filter_extreme_values == false {
502
502
  gene_infos.push(GeneInfo {
503
503
  rank_type: gene_counts_data.interquartile_range(),
504
- gene_symbol: gene_symbols[row].clone(),
504
+ gene_symbol: gene_names[row].clone(),
505
505
  });
506
506
  }
507
507
  }
@@ -536,28 +536,28 @@ fn cpm(
536
536
  }
537
537
 
538
538
  fn main() {
539
- // eprintln!("Starting gene variance calculation...");
539
+ // println!("Starting gene variance calculation...");
540
540
  let mut input = String::new();
541
541
  match io::stdin().read_line(&mut input) {
542
542
  // Accepting the piped input from nodejs (or command line from testing)
543
543
  Ok(_bytes_read) => {
544
- // eprintln!("Read {} bytes from stdin", _bytes_read);
545
- //println!("{} bytes read", bytes_read);
546
- //println!("{}", input);
544
+ // eprintln!("Read {} bytes from stdin", bytes_read);
545
+ // println!("{} bytes read", bytes_read);
546
+ // println!("{}", input);
547
547
  let input_json = json::parse(&input);
548
548
  match input_json {
549
549
  Ok(json_string) => {
550
- // eprintln!("Successfully parsed JSON input");
550
+ // println!("Successfully parsed JSON input");
551
551
  // let now = Instant::now();
552
552
  let samples_string_result = &json_string["samples"].to_owned();
553
553
  let samples_string;
554
554
  match samples_string_result.as_str() {
555
555
  Some(x) => {
556
556
  samples_string = x.to_string();
557
- // eprintln!("Samples: {}", samples_string);
557
+ // println!("Samples: {}", samples_string);
558
558
  }
559
559
  None => {
560
- eprintln!("ERROR: Samples not provided in JSON");
560
+ // eprintln!("ERROR: Samples not provided in JSON");
561
561
  println!(
562
562
  "{}",
563
563
  serde_json::json!({
@@ -583,14 +583,14 @@ fn main() {
583
583
  // );
584
584
  }
585
585
  None => {
586
- eprintln!("ERROR: File name missing in JSON");
587
- println!(
588
- "{}",
589
- serde_json::json!({
590
- "status": "error",
591
- "message": "File name is missing"
592
- })
593
- );
586
+ // eprintln!("ERROR: File name missing in JSON");
587
+ // println!(
588
+ // "{}",
589
+ // serde_json::json!({
590
+ // "status": "error",
591
+ // "message": "File name is missing"
592
+ // })
593
+ // );
594
594
  return;
595
595
  }
596
596
  }
@@ -613,14 +613,14 @@ fn main() {
613
613
  // eprintln!("Rank type: {}", rank_type);
614
614
  if rank_type != "var" && rank_type != "iqr" {
615
615
  // Check if any unknown method has been provided
616
- eprintln!("ERROR: Unknown rank method: {}", rank_type);
617
- println!(
618
- "{}",
619
- serde_json::json!({
620
- "status": "error",
621
- "message": format!("Unknown rank method: {}. Must be 'var' or 'iqr'", rank_type)
622
- })
623
- );
616
+ // eprintln!("ERROR: Unknown rank method: {}", rank_type);
617
+ // println!(
618
+ // "{}",
619
+ // serde_json::json!({
620
+ // "status": "error",
621
+ // "message": format!("Unknown rank method: {}. Must be 'var' or 'iqr'", rank_type)
622
+ // })
623
+ // );
624
624
  return;
625
625
  }
626
626
  let filter_extreme_values_result = &json_string["filter_extreme_values"];
@@ -629,7 +629,7 @@ fn main() {
629
629
  match filter_extreme_values_result.as_bool() {
630
630
  Some(x) => {
631
631
  filter_extreme_values = x;
632
- eprintln!("Filter extreme values: {}", filter_extreme_values);
632
+ // eprintln!("Filter extreme values: {}", filter_extreme_values);
633
633
  }
634
634
  None => {
635
635
  filter_extreme_values = true; // If filter_extreme_values field is missing, set it to true by default
@@ -648,7 +648,7 @@ fn main() {
648
648
  // eprintln!("Number of genes requested: {}", num_genes);
649
649
  }
650
650
  None => {
651
- eprintln!("ERROR: Number of genes to be given is missing");
651
+ // eprintln!("ERROR: Number of genes to be given is missing");
652
652
  println!(
653
653
  "{}",
654
654
  serde_json::json!({
@@ -668,7 +668,7 @@ fn main() {
668
668
  // eprintln!("Min count: {}", x);
669
669
  }
670
670
  None => {
671
- eprintln!("Min count not specified, will use default");
671
+ // eprintln!("Min count not specified, will use default");
672
672
  }
673
673
  }
674
674
 
@@ -680,7 +680,7 @@ fn main() {
680
680
  // eprintln!("Min total count: {}", x);
681
681
  }
682
682
  None => {
683
- eprintln!("Min total count not specified, will use default");
683
+ // eprintln!("Min total count not specified, will use default");
684
684
  }
685
685
  }
686
686
 
@@ -689,7 +689,7 @@ fn main() {
689
689
 
690
690
  // Choose the appropriate input function based on file type
691
691
  // eprintln!("Reading data from {} file: {}", file_type, file_name);
692
- let (input_matrix, gene_symbols) = if file_type == "hdf5" {
692
+ let (input_matrix, gene_names) = if file_type == "hdf5" {
693
693
  // eprintln!("Using HDF5 reader function...");
694
694
  match input_data_hdf5(&file_name, &samples_list) {
695
695
  Ok(result) => {
@@ -731,11 +731,11 @@ fn main() {
731
731
  // input_matrix.nrows(),
732
732
  // input_matrix.ncols()
733
733
  // );
734
- // eprintln!("Number of gene symbols: {}", gene_symbols.len());
735
- if !gene_symbols.is_empty() {
734
+ // eprintln!("Number of gene symbols: {}", gene_names.len());
735
+ if !gene_names.is_empty() {
736
736
  // eprintln!(
737
737
  // "First few gene symbols: {:?}",
738
- // &gene_symbols.iter().take(5).collect::<Vec<_>>()
738
+ // &gene_names.iter().take(5).collect::<Vec<_>>()
739
739
  // );
740
740
  }
741
741
 
@@ -749,7 +749,7 @@ fn main() {
749
749
  let gene_infos = match std::panic::catch_unwind(|| {
750
750
  calculate_variance(
751
751
  input_matrix,
752
- gene_symbols,
752
+ gene_names,
753
753
  samples_list.len() as f64,
754
754
  filter_extreme_values,
755
755
  rank_type.to_string(),
@@ -780,11 +780,11 @@ fn main() {
780
780
 
781
781
  // Check if we have enough genes for the requested output
782
782
  if gene_infos.len() < num_genes {
783
- eprintln!(
784
- "WARNING: Only {} genes found, but {} were requested",
785
- gene_infos.len(),
786
- num_genes
787
- );
783
+ // eprintln!(
784
+ // "WARNING: Only {} genes found, but {} were requested",
785
+ // gene_infos.len(),
786
+ // num_genes
787
+ // );
788
788
  }
789
789
 
790
790
  let actual_num_genes = std::cmp::min(num_genes, gene_infos.len());