1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164

use bio::stats::LogProb;

use model::sample::Observation;


pub type AlleleFreq = f64;


/// Variant calling model, taking purity and allele frequencies into account.
#[derive(Clone, Copy, Debug)]
pub struct LatentVariableModel {
    /// Purity of the case sample.
    purity: Option<LogProb>
}


impl LatentVariableModel {

    /// Create new model.
    pub fn new(purity: f64) -> Self {
        assert!(purity > 0.0 && purity <= 1.0);
        LatentVariableModel { purity: Some(LogProb(purity.ln())) }
    }

    pub fn with_single_sample() -> Self {
        LatentVariableModel { purity: None }
    }

    /// Impurity of the case sample (fraction of control cells in the case sample).
    fn impurity(&self) -> LogProb {
        self.purity.unwrap().ln_one_minus_exp()
    }

    /// Likelihood to observe a read given allele frequencies for case and control.
    fn likelihood_observation(&self,
                       observation: &Observation,
                       allele_freq_case: LogProb,
                       allele_freq_control: Option<LogProb>) -> LogProb {
        match (allele_freq_control, self.purity) {
            (Some(allele_freq_control), Some(purity)) => {
                // read comes from control sample and is correctly mapped
                let prob_control = self.impurity() +
                                   (allele_freq_control + observation.prob_alt).ln_add_exp(
                                         allele_freq_control.ln_one_minus_exp() + observation.prob_ref
                                   );
                // read comes from case sample and is correctly mapped
                let prob_case = purity +
                                (allele_freq_case + observation.prob_alt).ln_add_exp(
                                              allele_freq_case.ln_one_minus_exp() + observation.prob_ref
                                );
                // total probability
                let total = (observation.prob_mapping + prob_control.ln_add_exp(prob_case)).ln_add_exp(
                                  observation.prob_mapping.ln_one_minus_exp() + observation.prob_mismapped
                            );
                total
            },
            (None, purity) => {
                // no AF for control sample given
                if let Some(purity) = purity {
                    assert!(purity == LogProb::ln_one(), "no control allele frequency given but purity is not 1.0");
                }

                // read comes from case sample and is correctly mapped
                let prob_case = (allele_freq_case + observation.prob_alt).ln_add_exp(
                                              allele_freq_case.ln_one_minus_exp() + observation.prob_ref
                                );
                // total probability
                let total = (observation.prob_mapping + prob_case).ln_add_exp(
                                  observation.prob_mapping.ln_one_minus_exp() + observation.prob_mismapped
                            );
                total
            }
            (Some(_), None) => {
                panic!("control allele frequency given but purity no purity defined")
            }
        }
    }

    /// Likelihood to observe a pileup given allele frequencies for case and control.
    #[cfg_attr(feature="flame_it", flame)]
    pub fn likelihood_pileup(&self,
                             pileup: &[Observation],
                             allele_freq_case: f64,
                             allele_freq_control: Option<f64>) -> LogProb {
        let allele_freq_case = LogProb(allele_freq_case.ln());
        let allele_freq_control = allele_freq_control.map(|af| LogProb(af.ln()));
        // calculate product of per-read likelihoods in log space
        let likelihood = pileup.iter().fold(LogProb::ln_one(),
            |prob, obs| prob + self.likelihood_observation(obs, allele_freq_case, allele_freq_control));
        likelihood
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use model::sample::{Observation, Evidence};
    use itertools_num::linspace;
    use bio::stats::LogProb;

    #[test]
    fn test_likelihood_observation() {
        let model = LatentVariableModel::new(1.0);
        let observation = Observation{
            prob_mapping: LogProb::ln_one(),
            prob_alt: LogProb::ln_one(),
            prob_ref: LogProb::ln_zero(),
            prob_mismapped: LogProb::ln_one(),
            evidence: Evidence::dummy_alignment()
        };

        let lh = model.likelihood_observation(&observation, LogProb::ln_one(), Some(LogProb::ln_zero()));
        assert_relative_eq!(*lh, *LogProb::ln_one());

        let lh = model.likelihood_observation(&observation, LogProb::ln_zero(), Some(LogProb::ln_zero()));
        assert_relative_eq!(*lh, *LogProb::ln_zero());

        let lh = model.likelihood_observation(&observation, LogProb(0.5f64.ln()), Some(LogProb::ln_zero()));
        assert_relative_eq!(*lh, 0.5f64.ln());

        let lh = model.likelihood_observation(&observation, LogProb(0.5f64.ln()), Some(LogProb(0.5f64.ln())));
        assert_relative_eq!(*lh, 0.5f64.ln());

        let lh = model.likelihood_observation(&observation, LogProb(0.1f64.ln()), Some(LogProb::ln_zero()));
        assert_relative_eq!(*lh, 0.1f64.ln());

        // test with 50% purity
        let model = LatentVariableModel::new(0.5);

        let lh = model.likelihood_observation(&observation, LogProb::ln_zero(), Some(LogProb::ln_one()));
        assert_relative_eq!(*lh, 0.5f64.ln());
    }

    #[test]
    fn test_likelihood_pileup() {
        let model = LatentVariableModel::new(1.0);
        let mut observations = Vec::new();
        for _ in 0..5 {
            observations.push(Observation{
                prob_mapping: LogProb::ln_one(),
                prob_alt: LogProb::ln_one(),
                prob_ref: LogProb::ln_zero(),
                prob_mismapped: LogProb::ln_one(),
                evidence: Evidence::dummy_alignment()
            });
        }
        for _ in 0..5 {
            observations.push(Observation{
                prob_mapping: LogProb::ln_one(),
                prob_alt: LogProb::ln_zero(),
                prob_ref: LogProb::ln_one(),
                prob_mismapped: LogProb::ln_one(),
                evidence: Evidence::dummy_alignment()
            });
        }
        let lh = model.likelihood_pileup(&observations, 0.5, Some(0.0));
        for af in linspace(0.0, 1.0, 10) {
            if af != 0.5 {
                let l = model.likelihood_pileup(&observations, af, Some(0.0));
                assert!(lh > l);
            }
        }
    }
}