Segmentally Duplicated Regulatory Elements Undergo Human-Specific Rewiring.

1. package main
3. import (
4. "flag"
5. "fmt"
6. "github.com/vertgenlab/gonomics/fasta"
7. "github.com/vertgenlab/gonomics/motif"
8. "log"
9. )
11. type Settings struct {
12. InFile string
13. MatrixFile string
14. OutFile string
15. MatrixFileType string
16. PropMatch float64
17. Pseudocounts float64
18. OutputAsProportion bool
19. GcContent float64
20. }
22. func tfMatch(s Settings) {
23. records := fasta.Read(s.InFile)
24. fasta.AllToUpper(records)
25. motifs := motif.ReadJaspar(s.MatrixFile, s.MatrixFileType)
26. switch s.MatrixFileType {
27. case "Frequency":
28. motifs = motif.ReadJaspar(s.MatrixFile, "Frequency")
29. motifs = motif.PfmSliceToPpmSlice(motifs, s.Pseudocounts)
30. motifs = motif.PpmSliceToPwmSlice(motifs, s.GcContent)
31. case "Probability":
32. motifs = motif.ReadJaspar(s.MatrixFile, "Probability")
33. motifs = motif.PpmSliceToPwmSlice(motifs, s.GcContent)
34. case "Weight":
35. motifs = motif.ReadJaspar(s.MatrixFile, "Weight")
36. default:
37. log.Fatalf("Error. Unexpected motif file format. Options are 'Frequency', 'Probability', and 'Weight'.")
38. }
39. if s.PropMatch < 0 || s.PropMatch > 1 {
40. log.Fatalf("Error. PropMatch option should be a proportion, a value between 0 and 1.")
41. }
42. motif.RapidMatch(motifs, records, s.PropMatch, s.OutFile, s.OutputAsProportion)
43. }
45. func usage() {
46. fmt.Print(
47. "tfMatch - Genome-wide scanning of TFBS occurrences." +
48. "Input DNA sequences must be upper case.\n" +
49. "Usage:\n" +
50. "tfMatch input.fa matrices.pfm output.bed\n" +
51. "options:\n")
52. flag.PrintDefaults()
53. }
55. func main() {
56. var expectedNumArgs int = 3
57. var propMatch *float64 = flag.Float64("propMatch", 0.8, "Specifies the minimum motif score (as a proportion of the consensus sequence score) required for a match to be retained in the output.")
58. var matrixFileType *string = flag.String("matrixFileType", "Frequency", "Specify the type of position matrix file. Can be 'Frequency', 'Probability', or 'Weight'.")
59. var pfmPseudocounts *float64 = flag.Float64("pfmPseudocounts", 0.1, "If a Position Frequency Matrix is provided, this pseudocount value will be applied when converting to a PWM.")
60. var outputAsProportion *bool = flag.Bool("outputAsProportion", false, "Display the output motif scores as proportions of the consensus score. Motif difference score will thus be a change in consensus proportion.")
61. var gcContent *float64 = flag.Float64("gcContent", 0.5, "Set the expected GC content of the target sequence.")
63. flag.Usage = usage
64. log.SetFlags(log.Ldate | log.Ltime | log.Lshortfile)
65. flag.Parse()
67. if len(flag.Args()) != expectedNumArgs {
68. flag.Usage()
69. log.Fatalf("Error: expecting %d arguments, but got %d\n", expectedNumArgs, len(flag.Args()))
70. }
72. inFile := flag.Arg(0)
73. matrixFile := flag.Arg(1)
74. outFile := flag.Arg(2)
76. s := Settings{
77. InFile: inFile,
78. MatrixFile: matrixFile,
79. OutFile: outFile,
80. MatrixFileType: *matrixFileType,
81. PropMatch: *propMatch,
82. Pseudocounts: *pfmPseudocounts,
83. OutputAsProportion: *outputAsProportion,
84. GcContent: *gcContent,
85. }
87. tfMatch(s)
88. }

tfMatch.go at commit bd66b49, under BSD-3-Clause · at the source