Methods for the diagnosis and prognosis of melanoma from topical skin swabs

A system for predicting the presence of melanoma, comprising:

• a memory and at least one processor coupled to the memory, wherein the at least one processor is configured to:
• perform molecular profiling of at least one of genomic and transcriptomic material extracted from a skin swab sample collected from a lesion of a patient;
• wherein the performing of the molecular profiling comprises generating, with a sequencing machine, one or more computer files in a text-based format representing biological sequence data of the at least one of the genomic and transcriptomic material;
• and
• wherein the one or more computer files comprise at least one biological sequence;
• extract one or more biological features from the molecular profiling;
• wherein, in extracting, the at least one processor is configured to;
• compare the at least one biological sequence from the one or more computer files to a plurality of reference sequences to determine one or more differences between the at least one biological sequence and the plurality of reference sequences;
• and
• for at least one genomic material sample, input the at least one biological sequence from the one or more computer files to a rapid annotation component, and receive one or more microbial population variances from the rapid annotation component;
• wherein the plurality of reference sequences correspond to a plurality of reference skin swab samples collected from a plurality of reference patients;
• and
• evaluate, with a machine learning classifier, the one or more extracted biological features to generate a prognosis protocol comprising:
• (i) a score for the skin swab sample associated to a risk of cancer;
• (ii) a prediction whether the lesion of the patient is cancerous based at least in part on the score;
• and (iii) a computation of survival differences between groups of patients with different machine learning classifier outcomes;
• wherein, in evaluating, the at least one processor is configured to:
• generate a prediction model from a training data set comprising biological sequence samples respectively corresponding to one or more known labels of one of a cancerous group and a non-cancerous group;
• input the one or more differences and the one or more microbial population variances to the prediction model;
• and
• use the prediction model to generate the prognosis protocol including in part a determination whether the one or more differences and the one or more microbial population variances are indicative of cancer.

A computer program product for image analysis, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to perform a method comprising:

• performing molecular profiling of at least one of genomic and transcriptomic material extracted from a skin swab sample collected from a lesion of a patient;
• wherein the performing of the molecular profiling comprises generating, with a sequencing machine, one or more computer files in a text-based format representing biological sequence data of the at least one of the genomic and transcriptomic material;
• and
• wherein the one or more computer files comprise at least one biological sequence;
• extracting one or more biological features from the molecular profiling;
• wherein the extracting comprises:
• comparing the at least one biological sequence from the one or more computer files to a plurality of reference sequences to determine one or more differences between the at least one biological sequence and the plurality of reference sequences;
• and
• for at least one genomic material sample, inputting the at least one biological sequence from the one or more computer files to a rapid annotation component, and receiving one or more microbial population variances from the rapid annotation component;
• wherein the plurality of reference sequences correspond to a plurality of reference skin swab samples collected from a plurality of reference patients;
• and
• evaluating, with a machine learning classifier, the one or more extracted biological features to generate a prognosis protocol comprising:
• (i) a score for the skin swab sample associated to a risk of cancer;
• (ii) a prediction whether the lesion of the patient is cancerous based at least in part on the score;
• and (iii) a computation of survival differences between groups of patients with different machine learning classifier outcomes;
• wherein the evaluating comprises:
• generating a prediction model from a training data set comprising biological sequence samples respectively corresponding to one or more known labels of one of a cancerous group and a non-cancerous group;
• inputting the one or more differences and the one or more microbial population variances to the prediction model;
• and
• using the prediction model to generate the prognosis protocol including in part a determination whether the one or more differences and the one or more microbial population variances are indicative of cancer.